Comparative performance of verbal autopsy methods in identifying causes of adult mortality: A case study in India.

Background & objectives: Cause of death assignment from verbal autopsy (VA) questionnaires is conventionally accomplished through physician review. However, since recently, computer softwares have been developed to assign the cause of death. The present study evaluated the performance of computer software in assigning the cause of death from the VA, as compared to physician review. Methods: VA of 600 adult deaths was conducted using open- and close-ended questionnaires in Nandpur Kalour Block of Punjab, India. Entire VA forms were used by two physicians independently to assign the cause of death using the International Statistical Classification of Diseases and Related Health Problems (ICD)-10 codes. In case of disagreement between them, reconciliation was done, and in cases of persistent disagreements finally, adjudication was done by a third physician. InterVA-4-generated causes from close-ended questionnaires were compared using Kappa statistics with causes assigned by physicians using a questionnaire having both open- and close-ended questions. At the population level, Cause-Specific Mortality Fraction (CSMF) accuracy and P-value from McNemar's paired Chi-square were calculated. CSMF accuracy indicates the absolute deviation of a set of proportions of causes of death out of the total number of deaths between the two methods. Results: The overall agreement between InterVA-4 and physician coding was 'fair' (κ=0.42; 95% confidence interval 0.38, 0.46). CSMF accuracy was found to be 0.71. The differences in proportions from the two methods were statistically different as per McNemar's paired Chi-square test for ischaemic heart diseases, liver cirrhosis and maternal deaths. Interpretation & conclusions: In comparison to physicians, assignment of causes of death by InterVA- 4 was only 'fair'. Hence, it may be appropriate to continue with physician review as the optimal option available in the current scenario.

Assigning cause of maternal death: a comparison of findings by a facility-based review team, an expert panel using the new ICD-MM cause classification and a computer-based program (InterVA-4).

OBJECTIVE: To compare methodology used to assign cause of and factors contributing to maternal death. DESIGN: Reproductive Age Mortality Study. SETTING: Malawi. POPULATION: Maternal deaths among women of reproductive age. METHODS: We compared cause of death as assigned by a facility-based maternal death review team, an expert panel using the International Classification of Disease, 10th revision (ICD-10) cause classification for deaths during pregnancy, childbirth and the puerperium (ICD-MM) and a computer-based probabilistic program (InterVA-4). MAIN OUTCOME MEASURES: Number and cause of maternal deaths. RESULTS: The majority of maternal deaths occurred at a health facility (94/151; 62.3%). The estimated maternal mortality ratio was 363 per 100 000 live births (95% CI 307-425). There was poor agreement between cause of death assigned by a facility-based maternal death review team and an expert panel (κ = 0.37, 86 maternal deaths). The review team considered 36% of maternal deaths to be indirect and caused by non-obstetric complications (ICD-MM Group 7) whereas the expert panel considered only 17.4% to be indirect maternal deaths with 33.7% due to obstetric haemorrhage (ICD-MM Group 3). The review team incorrectly assigned a contributing condition rather than cause of death in up to 15.1% of cases. Agreement between the expert panel and InterVA-4 regarding cause of death was good (κ = 0.66, 151 maternal deaths). However, contributing conditions are not identified by InterVA-4. CONCLUSIONS: Training in the use of ICD-MM is needed for healthcare providers conducting maternal death reviews to be able to correctly assign underlying cause of death and contributing factors. Such information can help to identify what improvements in quality of care are needed. TWEETABLE ABSTRACT: For maternal deaths assigning cause of death is best done by an expert panel and helps to identify where quality of care needs to be improved.

Cause-specific mortality in the Kombewa health and demographic surveillance systems site, rural Western Kenya from 2011-2015.

BACKGROUND: The vast majority of deaths in the health and Kombewa demographic surveillance system (HDSS) study area are not registered and reported through official systems of vital registration. As a result, few data are available regarding causes of death in this population. OBJECTIVES: To describe causes of death among residents of all ages in the Kombewa HDSS, located in rural Western Kenya. METHODS: Verbal autopsy (VA) interviews at the site were conducted using the modified 2007 and later 2012 standardized WHO questionnaires. Assignment of causes of death was made using the InterVA-4 model version 4.02. Cox regression model, adjusted for sex, was built to evaluate the influence of age on mortality. RESULTS: There were a total of 5196 deaths recorded between 2011 and 2015 at the site. VA interviews were successfully completed for 3903 of these deaths (75.1%). Mortality rates were highest among neonates HR = 38.54 (<0.001) and among Infants HR = 2.07 (<0.006) in the Kombewa HDSS. Among those deaths in which VA was performed, the top causes of death were HIV/AIDS (12.6%), Malaria (10.3%), Pneumonia (10.1%), Acute abdomen (7.0%), Stroke (5.2%) and TB (4.9%) for the whole population in general. Stroke, acute abdomen heart diseases and Pneumonia were common causes of death (CODs) among the elderly over the age of 65. CONCLUSIONS: The analysis established the main CODs among people of all ages within the area served by the Kombewa HDSS. We hope that information generated from this study will help better address preventable deaths in the surveyed community as well as help mitigate negative health impacts in other rural communities throughout the Western Kenya region.

Causes of death in the Taabo health and demographic surveillance system, Côte d'Ivoire, from 2009 to 2011.

BACKGROUND: Current vital statistics from governmental institutions in Côte d'Ivoire are incomplete. This problem is particularly notable for remote rural areas that have limited access to the health system. OBJECTIVE: To record all deaths from 2009 to 2011 and to identify the leading causes of death in the Taabo health and demographic surveillance system (HDSS) in south-central Côte d'Ivoire. DESIGN: Deaths recorded in the first 3 years of operation of the Taabo HDSS were investigated by verbal autopsy (VA), using the InterVA-4 model. InterVA-4 is based on the World Health Organization 2012 VA tool in terms of input indicators and categories of causes of death. RESULTS: Overall, 948 deaths were recorded, of which 236 (24.9%) had incomplete VA data. Among the 712 deaths analyzed, communicable diseases represented the leading causes (58.9%), with most deaths attributed to malaria (n=129), acute respiratory tract infections (n=110), HIV/AIDS (n=80), and pulmonary tuberculosis (n=46). Non-communicable diseases accounted for 18.9% of the deaths and included mainly acute abdomen (n=38), unspecified cardiac diseases (n=15), and digestive neoplasms (n=13). Maternal and neonatal conditions accounted for 8.3% of deaths, primarily pneumonia (n=19) and birth asphyxia (n=16) in newborns. Among the 3.8% of deaths linked to trauma and injury, the main causes were assault (n=6), accidental drowning (n=4), contact with venomous plants/animals (n=4), and traffic-related accidents (n=4). No clear causes were determined in 10.0% of the analyzed deaths. CONCLUSIONS: Communicable diseases remain the predominant cause of death in rural Côte d'Ivoire. Based on these findings, measures are now being implemented in the Taabo HDSS. It will be interesting to monitor patterns of mortality and causes of death in the face of rapid demographic and epidemiological transitions in this part of West Africa.

Causes of death among persons of all ages within the Kilifi Health and Demographic Surveillance System, Kenya, determined from verbal autopsies interpreted using the InterVA-4 model.

BACKGROUND: The vast majority of deaths in the Kilifi study area are not recorded through official systems of vital registration. As a result, few data are available regarding causes of death in this population. OBJECTIVE: To describe the causes of death (CODs) among residents of all ages within the Kilifi Health and Demographic Surveillance System (KHDSS) on the coast of Kenya. DESIGN: Verbal autopsies (VAs) were conducted using the 2007 World Health Organization (WHO) standard VA questionnaires, and VA data further transformed to align with the 2012 WHO VA instrument. CODs were then determined using the InterVA-4 computer-based probabilistic model. RESULTS: Five thousand one hundred and eighty seven deaths were recorded between January 2008 and December 2011. VA interviews were completed for 4,460 (86%) deaths. Neonatal pneumonia and birth asphyxia were the main CODs in neonates; pneumonia and malaria were the main CODs among infants and children aged 1-4, respectively, while HIV/AIDS was the main COD for adult women of reproductive age. Road traffic accidents were more commonly observed among men than women. Stroke and neoplasms were common CODs among the elderly over the age of 65. CONCLUSIONS: We have established the main CODs among people of all ages within the area served by the KHDSS on the coast of Kenya using the 2007 WHO VA questionnaire coded using InterVA-4. We hope that our data will allow local health planners to estimate the burden of various diseases and to allocate their limited resources more appropriately.

Estimating cause of adult (15+ years) death using InterVA-4 in a rural district of southern Ghana.

BACKGROUND: Data needed to estimate causes of death and the pattern of these deaths are scarce in sub-Saharan Africa. Such data are very important for targeting, monitoring, and evaluating health interventions. OBJECTIVE: To estimate the mortality rate and determine causes of death among adults (aged 15 years and older) in a rural district of southern Ghana, using the InterVA-4 model. DESIGN: Data used were generated from verbal autopsies conducted for registered adult members of the Dodowa Health and Demographic Surveillance System who died between 2006 and 2010. The InterVA-4 model was used to assign the cause of death. RESULTS: Overall, the mortality rate for the period under review was 7.5/1,000 person-years (py) for the general population and 10.4/1,000 py for those aged 15 and older. The leading cause of death was communicable diseases (CDs), with a malaria-specific mortality rate of 1.06/1,000 py. Pulmonary tuberculosis (TB)-specific mortality rate was the next highest (1.01/1,000 py). HIV/AIDS attributed deaths were lower among males than females. Non-communicable diseases (NCDs) contributed to 28.3% of the deaths with cause-specific mortality rate of 2.93/1,000 py. Stroke topped the list with cause-specific mortality rate of 0.69/1,000 py. As expected, young males (15-49 years) contributed to more road traffic accident (RTA) deaths; they had a lower RTA cause-specific mortality rate than older males (50-64 years). CONCLUSIONS: Data indicate that CDs (e.g. malaria and TB) remain the major cause of death with NCDs (e.g. stroke) following closely behind. Verbal autopsy data can provide the causes of mortality in poorly resourced settings where access to timely and accurate data is scarce.

Cause of death during 2009-2012, using a probabilistic model (InterVA-4): an experience from Ballabgarh Health and Demographic Surveillance System in India.

OBJECTIVES: The present study aimed to estimate the age and cause-specific mortality in Ballabgarh Health and Demographic Surveillance System (HDSS) site for the years 2009 to 2012, using a probabilistic model (InterVA-4). METHODS: All Deaths in Ballabgarh HDSS from January 1, 2009, to December 31, 2012, were included in the study. InterVA-4 model (version 4.02) was used for assigning cause of death (COD). Data from the verbal autopsy (VA) tool were extracted and processed with the InterVA-4 model. Cause-specific mortality rate (CSMR) per 1,000 person-years was calculated. RESULTS: A total of 2,459 deaths occurred in the HDSS during the year 2009 to 2012. Among them, 2,174 (88.4%) valid VA interviews were conducted. Crude death rate ranged from 7.1 (2009) to 6.4 (2012) per 1,000 population. The CSMR per 1,000 person-years over the years (2009-2012) for non-communicable diseases, communicable diseases, trauma, neoplasm, and maternal and neonatal diseases were 1.78, 1.68, 0.68, 0.49, and 0.48, respectively. The most common causes of death among children, adults, and the elderly were infectious diseases, trauma, and non-communicable diseases, respectively. CONCLUSIONS: Overall, non-communicable diseases constituted the largest proportion of mortality, whereas trauma was the most common COD among adults at Ballabgarh HDSS. Policy-makers ought to focus on prevention of premature CODs, especially prevention of infectious diseases in children, and intentional self-harm and road traffic accidents in the adult population.