Henry D. Kalter

Population Health Metrics Research Consortium gold standard verbal autopsy validation study: design, implementation, and development of analysis datasets

BackgroundVerbal autopsy methods are critically important for evaluating the leading causes of death in populations without adequate vital registration systems. With a myriad of analytical and data collection approaches, it is essential to create a high quality validation dataset from different populations to evaluate comparative method performance and make recommendations for future verbal autopsy implementation. This study was undertaken to compile a set of strictly defined gold standard deaths for which verbal autopsies were collected to validate the accuracy of different methods of verbal autopsy cause of death assignment.MethodsData collection was implemented in six sites in four countries: Andhra Pradesh, India; Bohol, Philippines; Dar es Salaam, Tanzania; Mexico City, Mexico; Pemba Island, Tanzania; and Uttar Pradesh, India. The Population Health Metrics Research Consortium (PHMRC) developed stringent diagnostic criteria including laboratory, pathology, and medical imaging findings to identify gold standard deaths in health facilities as well as an enhanced verbal autopsy instrument based on World Health Organization (WHO) standards. A cause list was constructed based on the WHO Global Burden of Disease estimates of the leading causes of death, potential to identify unique signs and symptoms, and the likely existence of sufficient medical technology to ascertain gold standard cases. Blinded verbal autopsies were collected on all gold standard deaths.ResultsOver 12,000 verbal autopsies on deaths with gold standard diagnoses were collected (7,836 adults, 2,075 children, 1,629 neonates, and 1,002 stillbirths). Difficulties in finding sufficient cases to meet gold standard criteria as well as problems with misclassification for certain causes meant that the target list of causes for analysis was reduced to 34 for adults, 21 for children, and 10 for neonates, excluding stillbirths. To ensure strict independence for the validation of methods and assessment of comparative performance, 500 test-train datasets were created from the universe of cases, covering a range of cause-specific compositions.ConclusionsThis unique, robust validation dataset will allow scholars to evaluate the performance of different verbal autopsy analytic methods as well as instrument design. This dataset can be used to inform the implementation of verbal autopsies to more reliably ascertain cause of death in national health information systems.

The role of ethanol abuse in the etiology of heroin-related death.

Toxicology analyses and other forensic science data were used to examine the mechanisms through which ethanol increased the risk for death caused by injected street preparations of heroin. The authors studied 505 victims of fatal heroin overdose and compared subjects who had concentrations of blood ethanol greater than 1000 mg/L (n = 306) with those who had concentrations less than, or equal to 1000 mg/L (n = 199). We found significant negative correlations between concentrations of ethanol and morphine (a heroin metabolite) in blood (R2 = 0.11, P = 0.0001 for log10-transformed variables) as well as between concentrations of blood ethanol and bile morphine (R2 = 0.16, P = 0.0001 for log10 bile morphine versus blood morphine). Toxicologic evidence of infrequent heroin use was more common in decedents with blood ethanol concentrations greater than 1000 mg/L than in those with lower concentrations. Our data suggest that ethanol enhances the acute toxicity of heroin, and that ethanol use indirectly influences fatal overdose through its association with infrequent (nonaddictive) heroin use and thus with reduced tolerance to the acute toxic effects of heroin.

Using verbal autopsy to measure causes of death: the comparative performance of existing methods

BackgroundMonitoring progress with disease and injury reduction in many populations will require widespread use of verbal autopsy (VA). Multiple methods have been developed for assigning cause of death from a VA but their application is restricted by uncertainty about their reliability.MethodsWe investigated the validity of five automated VA methods for assigning cause of death: InterVA-4, Random Forest (RF), Simplified Symptom Pattern (SSP), Tariff method (Tariff), and King-Lu (KL), in addition to physician review of VA forms (PCVA), based on 12,535 cases from diverse populations for which the true cause of death had been reliably established. For adults, children, neonates and stillbirths, performance was assessed separately for individuals using sensitivity, specificity, Kappa, and chance-corrected concordance (CCC) and for populations using cause specific mortality fraction (CSMF) accuracy, with and without additional diagnostic information from prior contact with health services. A total of 500 train-test splits were used to ensure that results are robust to variation in the underlying cause of death distribution.ResultsThree automated diagnostic methods, Tariff, SSP, and RF, but not InterVA-4, performed better than physician review in all age groups, study sites, and for the majority of causes of death studied. For adults, CSMF accuracy ranged from 0.764 to 0.770, compared with 0.680 for PCVA and 0.625 for InterVA; CCC varied from 49.2% to 54.1%, compared with 42.2% for PCVA, and 23.8% for InterVA. For children, CSMF accuracy was 0.783 for Tariff, 0.678 for PCVA, and 0.520 for InterVA; CCC was 52.5% for Tariff, 44.5% for PCVA, and 30.3% for InterVA. For neonates, CSMF accuracy was 0.817 for Tariff, 0.719 for PCVA, and 0.629 for InterVA; CCC varied from 47.3% to 50.3% for the three automated methods, 29.3% for PCVA, and 19.4% for InterVA. The method with the highest sensitivity for a specific cause varied by cause.ConclusionsPhysician review of verbal autopsy questionnaires is less accurate than automated methods in determining both individual and population causes of death. Overall, Tariff performs as well or better than other methods and should be widely applied in routine mortality surveillance systems with poor cause of death certification practices.

Ascertaining causes of neonatal deaths using verbal autopsy: current methods and challenges

Objective:‘Verbal autopsy’ (VA) is used to ascertain cause of death in countries where vital registration systems are lacking. Current VA methods for neonatal deaths vary widely and suffer from several limitations. We aimed to: (1) review current neonatal VA methods, (2) identify gaps and limitations, (3) illustrate some limitations using VA data and (4) identify new approaches in methodology and analysis.Study Design:Rolling techniques and database search terms were used to identify articles that described neonatal VA administration, validation and cause of death assignment.Result:Current VA interviews include open and close-ended modules and are administered by trained interviewers. Causes of death are determined using physician review and/or computer algorithms for various neonatal causes of death. Challenges include lack of a standardized VA instrument and administration of methods, difficulty in identifying gold standards for validation studies, lack of validated algorithms for causes of death, poor existing algorithms, lack of standardized death classification terminology and the use of hierarchy to assign causes of death. Newer probabilistic methods of analysis such as Bayes Theorem or the Symptom Pattern method may improve accuracy for cause of death estimation and alleviate some of the challenges with traditional physician and algorithmic approaches, although additional research is needed.Conclusion:Given the continued reliance on VA to determine cause of death in settings with inadequate registration systems, it is important to understand the gaps in current VA methods and explore how methods can be improved to accurately reflect neonatal disease burden in the global community.

Potential implications of the integrated management of childhood illness (IMCI) for hospital referral and pharmaceutical usage in western Uganda

Summary The integrated management of childhood illness approach (IMCI) is currently being implemented by a number of countries worldwide. This is the second report from a study in western Uganda comparing the assessment and classification of disease by medical assistants using the IMCI algorithm with that of hospital‐based general medical officers, who used their clinical judgement to assess and provide treatment. Treatment prescribed by the hospital medical officers was compared to that indicated by IMCI disease classifications. The study population comprised 1226 children aged 2–59 months. Medical assistants had some difficulty in completing the IMCI assessment, leading to incorrect classification of findings in 138 of 1086 completed forms (13%). If their classifications had been used to decide on hospital referral, 37 children who met IMCI criteria for referral would have been sent home. Consultations took on average 7.2 min, longer than usual for several African countries. Use of the IMCI guidelines would have referred 16.2% of children to hospital, compared with 22% referred by the medical officers. Use of IMCI could have reduced the cost of medication to US

Diagnosis and management of febrile children using the WHO/UNICEF guidelines for IMCI in Dhaka, Bangladesh

0.17 per child compared to the treatment cost of US

Prospective community‐based cluster census and case‐control study of stillbirths and neonatal deaths in the West Bank and Gaza Strip

0.82 as prescribed by medical officers. Medical officers prescribed both a greater number and a greater variety of drugs than indicated by the IMCI algorithm. Compared to the present management of sick children by medical officers at Kabarole district hospital, using the IMCI algorithm would bring major changes in pharmaceutical use and referral practices. However, there is concern about the difficulty medical assistants had in using it, and the potential for longer consultation times.

Increased use of social autopsy is needed to improve maternal, neonatal and child health programmes in low-income countries

OBJECTIVE To determine whether the fever module in the WHO/UNICEF guidelines for the integrated management of childhood illness (IMCI) identifies children with bacterial infections in an area of low malaria prevalence. METHODS Physicians assessed a systematic sample of 669 sick children aged 2-59 months who presented to the outpatient department of Dhaka Shishu Hospital, Bangladesh. FINDINGS Had IMCI guidelines been used to evaluate the children, 78% of those with bacterial infections would have received antibiotics: the majority of children with meningitis (100%), pneumonia (95%), otitis media (95%) and urinary tract infection (83%); and 50% or less of children with bacteraemia (50%), dysentery (48%), and skin infections (30%). The current fever module identified only one additional case of meningitis. Children with bacteraemia were more likely to be febrile, feel hot, and have a history of fever than those with dysentery and skin infections. Fever combined with parental perception of fast breathing provided a more sensitive fever module for the detection of bacteraemia than the current IMCI module. CONCLUSIONS In an area of low malaria prevalence, the IMCI guidelines provide antibiotics to the majority of children with bacterial infections, but improvements in the fever module are possible.

Antimicrobial susceptibilities and serotype distribution of Streptococcus pneumoniae isolates from a Low socioeconomic area in Lima, Peru.

Obstetric complications and newborn illnesses amenable to basic medical interventions underlie most perinatal deaths. Yet, despite good access to maternal and newborn care in many transitional countries, perinatal mortality is often not monitored in these settings. The present study identified risk factors for perinatal death and the level and causes of stillbirths and neonatal deaths in the West Bank and Gaza Strip. Baseline and follow-up censuses with prospective monitoring of pregnant women and newborns from September 2001 to August 2002 were conducted in 83 randomly selected clusters of 300 households each. A total of 113 of 116 married women 15-49 years old with a stillbirth or neonatal death and 813 randomly selected women with a surviving neonate were interviewed, and obstetric and newborn care records of women with a stillbirth or neonatal death were abstracted. The perinatal and neonatal mortality rates, respectively, were 21.2 [95% confidence interval (CI) 16.5, 25.9] and 14.7 [95% CI 10.2, 19.2] per 1000 livebirths. The most common cause (27%) of 96 perinatal deaths was asphyxia alone (21) or with neonatal sepsis (5), while 18/49 (37%) early and 9/19 (47%) late neonatal deaths were from respiratory distress syndrome (12) or sepsis (9) alone or together (6). Constraint in care seeking, mainly by an Israeli checkpoint, occurred in 8% and 10%, respectively, of 112 pregnancies and labours and 31% of 16 neonates prior to perinatal or late neonatal death. Poor quality care for a complication associated with the death was identified among 40% and 20%, respectively, of 112 pregnancies and labour/deliveries and 43% of 68 neonates. (Correction added after online publication 5 June 2008: The denominators 112 pregnancies, labours, and labour/deliveries, and 16 and 68 neonates were included; and 9% of labours was corrected to 10%.) Risk factors for perinatal death as assessed by multivariable logistic regression included preterm delivery (odds ratio [OR] = 11.9, [95% CI 6.7, 21.2]), antepartum haemorrhage (OR = 5.6, [95% CI 1.5, 20.9]), any severe pregnancy complication (OR = 3.4, [95% CI 1.8, 6.6]), term delivery in a government hospital and having a labour and delivery complication (OR = 3.8, [95% CI 1.2, 12.0]), more than one delivery complication (OR = 4.4, [95% CI 1.8, 10.5]), mothers age >35 years (OR = 2.9, [95% CI 1.3, 6.8]) and primiparity in a full-term pregnancy (OR = 2.6, [1.1, 6.3]). Stillbirths are not officially reportable in the West Bank and Gaza Strip and this is the first time that perinatal mortality has been examined. Interventions to lower stillbirths and neonatal deaths should focus on improving the quality of medical care for important obstetric complications and newborn illnesses. Other transitional countries can draw lessons for their health care systems from these findings.

Social autopsy study identifies determinants of neonatal mortality in Doume, Nguelemendouka and Abong–Mbang health districts, Eastern Region of Cameroon

Although 2015 is only three years away, many countries are not on track to reach the United Nations Millennium Development Goals (MDGs), particularly goals 4 and 5, which call for reductions in child and maternal mortality rates. Targeted interventions are needed for rapid progress to take place. However, reliable estimates of the numbers, causes and determinants of death are needed to design the interventions. Such estimates are the basis of functional health information systems, which are the sources of data for the development and monitoring of evidence-based health policies and programmes.1 Countries unable to record the number of people who die or why they die cannot realize the full potential of their health systems.2 Nonetheless, many countries lack data on the social, behavioural and health systems determinants of child and maternal deaths. In some high-mortality settings, civil registration systems are so deficient that most deaths go unrecorded.3 Globally, fewer than one third of the 350 000 maternal deaths and 7 600 000 child deaths4 occurring annually are medically certified.5 The data needed to estimate cause-specific mortality rates can be obtained through alternative methods such as verbal autopsy, which consists of the use of standardized interview tools to question the caretakers of recently deceased persons about the symptoms that preceded the death.2Area-specific disease profiles generated from these data can help subnational and national health planners improve health resource allocation, as illustrated by the United Republic of Tanzania’s Essential Health Interventions Project.6 Yet despite the resources invested in developing verbal autopsy methods and tools, the non-biological factors contributing to a death cannot be determined from verbal autopsy data. This has led to the development of a complementary interview method known as “social autopsy”, which explores the social, behavioural and health systems determinants of maternal and child deaths.7 During social autopsy, the caretaker of the deceased is interviewed in detail regarding any preventive care received by the deceased, the diagnostic procedures followed, the type and timing of any treatment provided inside or outside the home, and any barriers encountered during care seeking.8 While verbal autopsy data can be used to prioritize health problems and evaluate health programme impact, social autopsy data, which focus on modifiable factors present in the home, community and health system, can inform policies and practices for increasing access to and use of preventive and curative services. The data afforded by social autopsy can create awareness that maternal and child deaths are preventable, empower communities to actively participate in interventions for reducing child and maternal deaths, and increase health programme responsiveness and accountability.9 Despite its importance, social autopsy has not been widely practiced and still lacks standard methods for data collection and analysis. Two key models, known as Pathway to Survival10 and Three Delays,11 have been used to organize the care-seeking data generated by social autopsy.8,9,12 These models should guide future social autopsy studies, whose aim should be to maximize social autopsy’s awareness-raising potential through participatory data sharing and development of interventions.7 The World Health Organization’s Child Health Epidemiology Reference Group has been further developing and testing the Pathway to Survival and Three Delays analytical models in several low-income countries. The INDEPTH Network has also been developing its own social autopsy tools. Both groups have identified similar challenges, primarily the need to reduce the number of interview questions to make data collection faster, and to develop a suitable strategy for data analysis. The standard social autopsy tool intended for use in combination with verbal autopsy is already under review to try to reduce the number of questions, but more resources should be invested in trying to simplify and standardize social autopsy methods and in encouraging the use of social autopsy, which is a new and still unfamiliar method. Resources should also be invested in bringing together interested parties with the aim of reaching consensus on standardized formats and computer-aided data analysis plans. All of these measures will enhance the visibility of social autopsy and promote routine collection and use of quality social autopsy data. Social autopsy instruments should be widely disseminated, along with tools for country adaptation, fieldworker training and community data sharing, and for integration with population-wide surveys. Tools of this type are direly needed for low-income countries to achieve the MDGs established for the reduction of child and maternal mortality.