Counting the Dead, Making the Dead Count (including everyone) throughHealthy Sierra Leone — HEAL-SL

Abstract

The specifics of mortality—who, where, when, and why—are pillars of public health: vital statistics to understand population health, set priorities, and monitor progress. Yet in sub-Saharan Africa, few nationally representative mortality data are collected, and there is a reliance on estimates from global models. Healthy Sierra Leone (HEAL-SL) has produced the first reliable cause-of-death data for the country for 2018‒2023, based on ‘verbal autopsies’ (VAs) for a representative 5% of the population. Over 50 trained African survey staff cover 700 areas, engaging with local communities and implementing fully electronic, near real-time surveys. A central team of physicians examines each record to determine cause of death.

Heretofore undocumented, malaria kills one-quarter of the Sierra Leoneans who die prematurely. Maternal deaths are only half as frequent as stated in the models. Deaths from coronavirus 2019 (COVID-19) are far lower than expected. HEAL-SL is about counting the dead to help the living.

Introduction

The implementing partners of HEAL-SL are the School of Community Health Sciences at Njala University (NU) in Bo, Sierra Leone, the Centre for Global Health Research (CGHR) at the University of Toronto (U of T) in Canada, and the Sierra Leone Ministry of Health and Sanitation (MoHS). NU Professor Rashid Ansumana oversees the fieldwork, and with a small senior local staff, is responsible for hiring and ensuring the proficiency of field staff, keeping the project on schedule, maintaining communications with the government and others, and overall management. CGHR is the main architect of the HEAL-SL sampling frame, establishing standard operating procedures, setting up the analytical framework, and collaborating on all analyses. The MoHS is the key user. The Bill and Melinda Gates Foundation (BMGF) and the U of T are the key funders.

Who should benefit?

Nothing is more fundamental to understanding the health of a population than its record of births and deaths. Yet in sub-Saharan Africa, few countries record even the fact of death for most of the population, and almost none record the circumstances leading to death or its causes.

Every country in Europe, North America, and much of the rest of the world collects these data as an essential function of government. Countries without adequate vital statistics—including until recently Sierra Leone—rely on modelled data from United Nations agencies on the numbers and causes of deaths, which are a poor substitute for directly collected information. We view the absence of cause-of-death information as an unmet need for the entire population.

In 2017, when the idea for HEAL-SL was born, only 25% of deaths there were reported officially, without recording the causes. Professor Prabhat Jha of the U of T is the architect and initial implementer of the Million Death Study (MDS), an audacious plan to establish a nationally representative cause-of-death reporting system for India in the early 2000s. Recognizing that 100% coverage would not be immediately feasible, the MDS sampled about 1% of the population, covering over 1 million homes across the country. Because deaths occurred mainly outside of health facilities, the MDS was based on VAs—structured conversations with next-of-kin in the months after a death occurred, with cause-of-death assigned using standard coding rules by expert physicians. After running for a decade as an academic‒public partnership, the MDS was taken over entirely by the government as the scaffolding of its mortality statistics system.

In 2017, Jha set his sights on establishing mortality-reporting capability in Africa, the region with the greatest unmet need. He found a willing partner in Ansumana, Dean of the School of Community Health Sciences at NU. Ansumana is a leader in the academic health community (who started a medical school at NU) with a strong relationship with the MoHS and a host of public-health trainees eager to work in the field.

Investing in vital statistics infrastructure in Sierra Leone was attractive to the BMGF, which has supported the first five years of HEAL-SL, to get it up and running and ready to hand over to the government and secure funding for the long term.

Engagement

The success of HEAL-SL is all about continual engagement. Once the partnership with Ansumana and NU had been established, the focus was on engaging the MoHS at national and district levels to buy into the idea that tracking and understanding births and deaths was a worthwhile investment. The value may seem obvious, but in one of the poorest countries in sub-Saharan Africa, scepticism is a constant, and the trade-offs of dealing with omnipresent health emergencies can be a high barrier. This is true even with external funding, as human resources are as scarce as dollars.

More essential, and often challenging, is the ongoing engagement with the Sierra Leonean population. The social mobilization (‘soc mob’) sessions preceding each round of data collection were originally intended to alert households to expect surveyors and explain what they would be asked about and why. This continues, but people also want to get something back for their time and cooperation. With robust results in hand, we are pilot testing using soc mobs to feed back to all communities—not just those in the sample—what was found and what they can do to prevent many of the premature deaths, using straightforward messages such as how to prevent disease and recognize the need for medical care for common problems, and avoiding traffic accidents and falls. ‘Report cards’ are being prepared with top-line messages and actions that can be taken at the chiefdom level. District Medical Officers (DMOs) will get different messages—for example, that they should check health centres for supplies to diagnose and treat malaria, ensure blood-pressure checks for adults, and make sure pregnant women have access to preventive and health-promoting services.

In October 2023, the 16 DMOs were brought together with academic experts to review the HEAL-SL results and develop action plans to address some of the leading causes of death across the age groups—stillbirths, malaria, road traffic accidents, and strokes. This involved not only health personnel but also the police, for guidance in addressing traffic deaths.

Turning plans into actions is proving slow, but progress is being made. One current focus is to promote HEAL-SL to an expanded circle of Sierra Leonean health advocates and researchers across the health spectrum to work with the data, which we have made publicly available at www.healsl.org. We are planning online and in-person seminars to introduce HEAL-SL and encourage widespread collaboration.

Research

HEAL-SL surveys a representative 5% of Sierra Leone’s population each year to determine who is born, who dies and from what causes, based on electronic VAs (e-VAs). It is a collaboration of the MoHS, NU, and U of T. It is designed to generate separate urban and rural district-level estimates of annual age-specific and cause-specific mortality rates. In December 2023, HEAL-SL released a mortality report for deaths from 2018 to 2023.

HEAL-SL was created entirely from scratch. The first step was to establish a sampling frame covering the whole of Sierra Leone. The sampling units are urban census blocks or villages known as enumeration areas (EAs). EAs in each district were randomly selected until about 5% of the population was included, resulting in 678 geographically dispersed EAs covering about 340,000 members of the population.

Before starting in each area, local district soc mob teams hold sensitization meetings with village leaders and community members. The surveys have two stages. First, surveyors enumerate the population in each house and identify deaths (in the initial phase, those occurring from 2018 to 2020). Second, surveyors in teams of four or five return to households reporting deaths and conduct e-VAs. Teams complete this work with quality assurance steps in one week for each EA.

Surveyors interview the families of decedents using the World Health Organization (WHO) VA standard tool, on electronic tablets, to gather details about the circumstances of each death. Each family is asked to describe in their own words what happened. Using all the information, two physicians trained in death certification and coding independently assign underlying causes of death using the 10th revision of the International Classification of Diseases (ICD-10) codes, with reconciliation of differences by a third senior physician. Coded deaths are grouped into 45 sets of causes, based on WHO guidance.

The results are normalized to the entire population by age group, urban or rural setting, geography, and social status. We found that almost two-thirds of Sierra Leoneans die prematurely, before age 70. We confirmed for the first time that malaria was the leading cause of death nationally, not only among children but also among adults. We found major discrepancies with the pre-HEAL-SL modelled mortality estimates for maternal mortality and many other specific causes. Challenges were faced daily, such as lack of electricity or impassable roads, mainly due to the poor infrastructure in a poor country. But our ultimate success in Sierra Leone means that ‘HEAL-you-name-the-country’ should be feasible almost anywhere.