2015/04/08 by Samuel J. Clark, Clark, Samuel J., Jon Wakefield +6
Mathematics · Medicine · #Applications (stat.AP) #COVID-19 epidemiological studies #Cause of death #Census #Census and Population Estimation #Cluster sampling #Computer science #Data collection #Data-Driven Disease Surveillance #Environmental health #FOS: Computer and information sciences #Geography #Lot quality assurance sampling #Mathematics #Medicine #Nonprobability sampling #Other Statistics (stat.OT) #Population #Public health #Public health surveillance #Sample (material) #Sampling (signal processing) #Socioeconomic status #Statistics #Survey sampling #Telecommunications #Verbal autopsy #stat.AP #stat.OT
paper · pdf · doi:10.48550/arxiv.1504.02124
Updated version including new simulation study with two-stage cluster and optimum allocation sampling
openalex publication_date 2015/04/08 · arxiv created 2017/05/26 · arxiv updated 2017/05/29 · openalex created_date 2025/10/10 · openalex updated_date 2026/08/05
Traditionally health statistics are derived from civil and/or vital\nregistration. Civil registration in low-income countries varies from partial\ncoverage to essentially nothing at all. Consequently the state of the art for\npublic health information in low-income countries is efforts to combine or\ntriangulate data from different sources to produce a more complete picture\nacross both time and space - data amalgamation. Data sources amenable to this\napproach include sample surveys, sample registration systems, health and\ndemographic surveillance systems, administrative records, census records,\nhealth facility records and others.\n We propose a new statistical framework for gathering health and population\ndata - Hyak - that leverages the benefits of sampling and longitudinal,\nprospective surveillance to create a cheap, accurate, sustainable monitoring\nplatform. Hyak has three fundamental components:\n 1) Data Amalgamation: a sampling and surveillance component that organizes\ntwo or more data collection systems to work together: a) data from HDSS with\nfrequent, intense, linked, prospective follow-up and b) data from sample\nsurveys conducted in large areas surrounding the Health and Demographic\nSurveillance System sites using informed sampling so as to capture as many\nevents as possible;\n 2) Cause of Death: verbal autopsy to characterize the distribution of deaths\nby cause at the population level; and\n 3) SES: measurement of socioeconomic status in order to characterize poverty\nand wealth.\n We conduct a simulation study of the informed sampling component of Hyak\nbased on the Agincourt HDSS site in South Africa. Compared to traditional\ncluster sampling, Hyak's informed sampling captures more deaths, and when\ncombined with an estimation model that includes spatial smoothing, produces\nestimates mortality that have lower variance and small bias.\n