[Paper Review] Hyak Mortality Monitoring System: Innovative Sampling and Estimation Methods - Proof of Concept by Simulation

Samuel J. Clark, Jon Wakefield+2 more|PubMed|Apr 8, 2015

Data-Driven Disease Surveillance16 references3 citations

TL;DR

This paper proposes Hyak, a novel statistical framework for low- and middle-income countries that integrates Health and Demographic Surveillance Systems (HDSS) with informed sampling from surrounding areas to improve mortality monitoring. Using simulation based on the Agincourt HDSS site, Hyak’s informed sampling strategy captures more deaths and produces mortality estimates with lower variance and small bias compared to traditional cluster sampling.

ABSTRACT

Traditionally health statistics are derived from civil and/or vital registration. Civil registration in low- to middle-income countries varies from partial coverage to essentially nothing at all. Consequently the state of the art for public health information in low- to middle-income countries is efforts to combine or triangulate data from different sources to produce a more complete picture across both time and space - <i>data amalgamation</i>. Data sources amenable to this approach include sample surveys, sample registration systems, health and demographic surveillance systems, administrative records, census records, health facility records and others. We propose a new statistical framework for gathering health and population data - Hyak - that leverages the benefits of sampling and longitudinal, prospective surveillance to create a cheap, accurate, sustainable monitoring platform. Hyak has three fundamental components: <i>Data amalgamation</i>: A sampling and surveillance component that organizes two or more data collection systems to work together: (1) data from HDSS with frequent, intense, linked, prospective follow-up and (2) data from sample surveys conducted in large areas surrounding the Health and Demographic Surveillance System (HDSS) sites using informed sampling so as to capture as many events as possible;<i>Cause of death</i>: Verbal autopsy to characterize the distribution of deaths by cause at the population level; and<i>Socioeconomic status (SES)</i>: Measurement of SES in order to characterize poverty and wealth. We conduct a simulation study of the informed sampling component of Hyak based on the Agincourt HDSS site in South Africa. Compared with traditional cluster sampling, Hyak's informed sampling captures more deaths, and when combined with an estimation model that includes spatial smoothing, produces estimates of both mortality counts and mortality rates that have lower variance and small bias.

Motivation & Objective

• Address the critical lack of reliable vital statistics in low- and middle-income countries due to weak or non-existent civil registration systems.
• Develop a sustainable, cost-effective platform for monitoring population health indicators, especially mortality rates and cause-specific mortality fractions (CSMFs).
• Overcome the limitations of traditional cluster sampling by leveraging longitudinal HDSS data and spatial modeling to optimize sampling efficiency.
• Create a scalable, integrated system that supports both public health surveillance and scientific research through combined HDSS and survey infrastructure.
• Enable precise, disaggregated monitoring of mortality trends by age, sex, cause, and socioeconomic status over time and space.

Proposed method

• Implement a data amalgamation framework combining HDSS data with sample surveys from surrounding areas using informed sampling strategies.
• Use verbal autopsy to determine cause-specific mortality fractions (CSMFs) at the population level.
• Incorporate socioeconomic status (SES) measurement to assess poverty and wealth distribution across populations.
• Apply spatial smoothing in estimation models to improve precision of mortality rate and count estimates.
• Utilize historical survey data (e.g., DHS) to build predictive surfaces for preferential sampling, informing where to sample next.
• Conduct a simulation study in the Agincourt HDSS region (South Africa) to compare informed sampling against traditional cluster sampling.

Experimental results

Research questions

• RQ1 Can informed sampling based on HDSS data capture more deaths than traditional cluster sampling in low-resource settings?
• RQ2 How does informed sampling combined with spatial smoothing affect the precision and bias of mortality rate and count estimates?
• RQ3 What is the optimal number and geographic distribution of HDSS sites needed to support effective informed sampling across a region?
• RQ4 How do sampling frame quality and update frequency affect the performance of the Hyak system?
• RQ5 To what extent can geostatistical models predict mortality in unobserved villages using data from sampled HDSS and surrounding areas?

Key findings

• Hyak’s informed sampling strategy captured significantly more deaths than traditional cluster sampling in the simulation study.
• Mortality rate and count estimates from Hyak demonstrated lower variance compared to traditional methods, with only small bias.
• The integration of spatial smoothing in the estimation model enhanced precision, particularly in areas with sparse data.
• The system demonstrated potential for producing reliable, disaggregated estimates of mortality by age, sex, cause, and socioeconomic status.
• Hyak is more cost-effective than traditional cluster surveys, offering greater information per dollar spent due to higher death capture efficiency.
• The framework supports scalable, sustainable monitoring by combining permanent HDSS infrastructure with periodic, targeted surveys.