RESUMO
BACKGROUND: A verbal autopsy (VA) is an interview conducted with the caregivers of someone who has recently died to describe the circumstances of the death. In recent years, several algorithmic methods have been developed to classify cause of death using VA data. The performance of one method-InSilicoVA-was evaluated in a study by Flaxman et al., published in BMC Medicine in 2018. The results of that study are different from those previously published by our group. METHODS: Based on the description of methods in the Flaxman et al. study, we attempt to replicate the analysis to understand why the published results differ from those of our previous work. RESULTS: We failed to reproduce the results published in Flaxman et al. Most of the discrepancies we find likely result from undocumented differences in data pre-processing, and/or values assigned to key parameters governing the behavior of the algorithm. CONCLUSION: This finding highlights the importance of making replication code available along with published results. All code necessary to replicate the work described here is freely available on GitHub.
Assuntos
Autopsia/métodos , Causas de Morte/tendências , Humanos , Projetos de Pesquisa , Estudos de Validação como AssuntoRESUMO
BACKGROUND: Opioid misuse and deaths are increasing in the United States. In 2017, Ohio had the second highest overdose rates in the US, with the city of Cincinnati experiencing a 50% rise in opioid overdoses since 2015. Understanding the temporal and geographic variation in overdose emergencies may help guide public policy responses to the opioid epidemic. METHODS AND FINDINGS: We used a publicly available data set of suspected heroin-related emergency calls (n = 6,246) to map overdose incidents to 280 census block groups in Cincinnati between August 1, 2015, and January 30, 2019. We used a Bayesian space-time Poisson regression model to examine the relationship between demographic and environmental characteristics and the number of calls within block groups. Higher numbers of heroin-related incidents were found to be associated with features of the built environment, including the proportion of parks (relative risk [RR] = 2.233; 95% credible interval [CI]: [1.075-4.643]), commercial (RR = 13.200; 95% CI: [4.584-38.169]), manufacturing (RR = 4.775; 95% CI: [1.958-11.683]), and downtown development zones (RR = 11.362; 95% CI: [3.796-34.015]). The number of suspected heroin-related emergency calls was also positively associated with the proportion of male population, the population aged 35-49 years, and distance to pharmacies and was negatively associated with the proportion aged 18-24 years, the proportion of the population with a bachelor's degree or higher, median household income, the number of fast food restaurants, distance to hospitals, and distance to opioid treatment programs. Significant spatial and temporal heterogeneity in the risks of incidents remained after adjusting for covariates. Limitations of this study include lack of information about the nature of incidents after dispatch, which may differ from the initial classification of being related to heroin, and lack of information on local policy changes and interventions. CONCLUSIONS: We identified areas with high numbers of reported heroin-related incidents and features of the built environment and demographic characteristics that are associated with these events in the city of Cincinnati. Publicly available information about opiate overdoses, combined with data on spatiotemporal risk factors, may help municipalities plan, implement, and target harm-reduction measures. In the US, more work is necessary to improve data availability in other cities and states and the compatibility of data from different sources in order to adequately measure and monitor the risk of overdose and inform health policies.
Assuntos
Overdose de Drogas/epidemiologia , Dependência de Heroína/epidemiologia , Teorema de Bayes , Bases de Dados Factuais , Serviços Médicos de Emergência/tendências , Serviço Hospitalar de Emergência/tendências , Feminino , Heroína/efeitos adversos , Humanos , Masculino , Ohio/epidemiologia , Fatores de Risco , Análise Espaço-Temporal , Transtornos Relacionados ao Uso de Substâncias/epidemiologia , Estados UnidosRESUMO
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.
Assuntos
Autopsia/métodos , Coleta de Dados/métodos , Médicos/normas , Adulto , Criança , Morte , Feminino , Humanos , Índia , MasculinoRESUMO
Verbal autopsy (VA) is a survey-based tool widely used to infer cause of death (COD) in regions without complete-coverage civil registration and vital statistics systems. In such settings, many deaths happen outside of medical facilities and are not officially documented by a medical professional. VA surveys, consisting of signs and symptoms reported by a person close to the decedent, are used to infer the COD for an individual, and to estimate and monitor the COD distribution in the population. Several classification algorithms have been developed and widely used to assign causes of death using VA data. However, the incompatibility between different idiosyncratic model implementations and required data structure makes it difficult to systematically apply and compare different methods. The openVA package provides the first standardized framework for analyzing VA data that is compatible with all openly available methods and data structure. It provides an open-source, R implementation of several most widely used VA methods. It supports different data input and output formats, and customizable information about the associations between causes and symptoms. The paper discusses the relevant algorithms, their implementations in R packages under the openVA suite, and demonstrates the pipeline of model fitting, summary, comparison, and visualization in the R environment.
RESUMO
Close contact between people is the primary route for transmission of SARS-CoV-2, the virus that causes coronavirus disease 2019 (COVID-19). We quantified interpersonal contact at the population level using mobile device geolocation data. We computed the frequency of contact (within 6 feet) between people in Connecticut during February 2020 to January 2021 and aggregated counts of contact events by area of residence. When incorporated into a SEIR-type model of COVID-19 transmission, the contact rate accurately predicted COVID-19 cases in Connecticut towns. Contact in Connecticut explains the initial wave of infections during March to April, the drop in cases during June to August, local outbreaks during August to September, broad statewide resurgence during September to December, and decline in January 2021. The transmission model fits COVID-19 transmission dynamics better using the contact rate than other mobility metrics. Contact rate data can help guide social distancing and testing resource allocation.
RESUMO
To support public health policymakers in Connecticut, we developed a flexible county-structured compartmental SEIR-type model of SARS-CoV-2 transmission and COVID-19 disease progression. Our goals were to provide projections of infections, hospitalizations, and deaths, and estimates of important features of disease transmission and clinical progression. In this paper, we outline the model design, implementation and calibration, and describe how projections and estimates were used to meet the changing requirements of policymakers and officials in Connecticut from March 2020 to February 2021. The approach takes advantage of our unique access to Connecticut public health surveillance and hospital data and our direct connection to state officials and policymakers. We calibrated this model to data on deaths and hospitalizations and developed a novel measure of close interpersonal contact frequency to capture changes in transmission risk over time and used multiple local data sources to infer dynamics of time-varying model inputs. Estimated epidemiologic features of the COVID-19 epidemic in Connecticut include the effective reproduction number, cumulative incidence of infection, infection hospitalization and fatality ratios, and the case detection ratio. We conclude with a discussion of the limitations inherent in predicting uncertain epidemic trajectories and lessons learned from one year of providing COVID-19 projections in Connecticut.
Assuntos
COVID-19 , Modelos Estatísticos , Pandemias , Vigilância em Saúde Pública/métodos , COVID-19/epidemiologia , COVID-19/transmissão , Connecticut/epidemiologia , Previsões , Humanos , Pandemias/prevenção & controle , Pandemias/estatística & dados numéricosRESUMO
To support public health policymakers in Connecticut, we developed a county-structured compartmental SEIR-type model of SARS-CoV-2 transmission and COVID-19 disease progression. Our goals were to provide projections of infections, hospitalizations, and deaths, as well as estimates of important features of disease transmission, public behavior, healthcare response, and clinical progression of disease. In this paper, we describe a transmission model developed to meet the changing requirements of public health policymakers and officials in Connecticut from March 2020 to February 2021. We outline the model design, implementation and calibration, and describe how projections and estimates were used to support decision-making in Connecticut throughout the first year of the pandemic. We calibrated this model to data on deaths and hospitalizations, developed a novel measure of close interpersonal contact frequency to capture changes in transmission risk over time and used multiple local data sources to infer dynamics of time-varying model inputs. Estimated time-varying epidemiologic features of the COVID-19 epidemic in Connecticut include the effective reproduction number, cumulative incidence of infection, infection hospitalization and fatality ratios, and the case detection ratio. We describe methodology for producing projections of epidemic evolution under uncertain future scenarios, as well as analytical tools for estimating epidemic features that are difficult to measure directly, such as cumulative incidence and the effects of non-pharmaceutical interventions. The approach takes advantage of our unique access to Connecticut public health surveillance and hospital data and our direct connection to state officials and policymakers. We conclude with a discussion of the limitations inherent in predicting uncertain epidemic trajectories and lessons learned from one year of providing COVID-19 projections in Connecticut.
RESUMO
Close contact between people is the primary route for transmission of SARS-CoV-2, the virus that causes coronavirus disease 2019 (COVID-19). We sought to quantify interpersonal contact at the population-level by using anonymized mobile device geolocation data. We computed the frequency of contact (within six feet) between people in Connecticut during February 2020 - January 2021. Then we aggregated counts of contact events by area of residence to obtain an estimate of the total intensity of interpersonal contact experienced by residents of each town for each day. When incorporated into a susceptible-exposed-infective-removed (SEIR) model of COVID-19 transmission, the contact rate accurately predicted COVID-19 cases in Connecticut towns during the timespan. The pattern of contact rate in Connecticut explains the large initial wave of infections during March-April, the subsequent drop in cases during June-August, local outbreaks during August-September, broad statewide resurgence during September-December, and decline in January 2021. Contact rate data can help guide public health messaging campaigns to encourage social distancing and in the allocation of testing resources to detect or prevent emerging local outbreaks more quickly than traditional case investigation. ONE SENTENCE SUMMARY: Close interpersonal contact measured using mobile device location data explains dynamics of COVID-19 transmission in Connecticut during the first year of the pandemic.
RESUMO
Learning dependence relationships among variables of mixed types provides insights in a variety of scientific settings and is a well-studied problem in statistics. Existing methods, however, typically rely on copious, high quality data to accurately learn associations. In this paper, we develop a method for scientific settings where learning dependence structure is essential, but data are sparse and have a high fraction of missing values. Specifically, our work is motivated by survey-based cause of death assessments known as verbal autopsies (VAs). We propose a Bayesian approach to characterize dependence relationships using a latent Gaussian graphical model that incorporates informative priors on the marginal distributions of the variables. We demonstrate such information can improve estimation of the dependence structure, especially in settings with little training data. We show that our method can be integrated into existing probabilistic cause-of-death assignment algorithms and improves model performance while recovering dependence patterns between symptoms that can inform efficient questionnaire design in future data collection.
RESUMO
The distribution of deaths by cause provides crucial information for public health planning, response and evaluation. About 60% of deaths globally are not registered or given a cause, limiting our ability to understand disease epidemiology. Verbal autopsy (VA) surveys are increasingly used in such settings to collect information on the signs, symptoms and medical history of people who have recently died. This article develops a novel Bayesian method for estimation of population distributions of deaths by cause using verbal autopsy data. The proposed approach is based on a multivariate probit model where associations among items in questionnaires are flexibly induced by latent factors. Using the Population Health Metrics Research Consortium labeled data that include both VA and medically certified causes of death, we assess performance of the proposed method. Further, we estimate important questionnaire items that are highly associated with causes of death. This framework provides insights that will simplify future data.
RESUMO
Bayesian graphical models are a useful tool for understanding dependence relationships among many variables, particularly in situations with external prior information. In high-dimensional settings, the space of possible graphs becomes enormous, rendering even state-of-the-art Bayesian stochastic search computationally infeasible. We propose a deterministic alternative to estimate Gaussian and Gaussian copula graphical models using an Expectation Conditional Maximization (ECM) algorithm, extending the EM approach from Bayesian variable selection to graphical model estimation. We show that the ECM approach enables fast posterior exploration under a sequence of mixture priors, and can incorporate multiple sources of information.
RESUMO
In regions without complete-coverage civil registration and vital statistics systems there is uncertainty about even the most basic demographic indicators. In such regions the majority of deaths occur outside hospitals and are not recorded. Worldwide, fewer than one-third of deaths are assigned a cause, with the least information available from the most impoverished nations. In populations like this, verbal autopsy (VA) is a commonly used tool to assess cause of death and estimate cause-specific mortality rates and the distribution of deaths by cause. VA uses an interview with caregivers of the decedent to elicit data describing the signs and symptoms leading up to the death. This paper develops a new statistical tool known as InSilicoVA to classify cause of death using information acquired through VA. InSilicoVA shares uncertainty between cause of death assignments for specific individuals and the distribution of deaths by cause across the population. Using side-by-side comparisons with both observed and simulated data, we demonstrate that InSilicoVA has distinct advantages compared to currently available methods.