Improving the estimation of the global burden of antimicrobial resistant infections

Direk Limmathurotsakul, Susanna Dunachie, Keiji Fukuda, et al.

Lancet Infect Dis Published:August 16, 2019 DOI:https://doi.org/10.1016/S1473-3099(19)30276-2

Summary

Estimating the global burden of disease from infections caused by pathogens that have acquired antimicrobial resistance (AMR) is essential for resource allocation and to inform AMR action plans at national and global levels. However, the scarcity of robust and accepted methods to determine burden is widely acknowledged. In this Personal View, we discuss the underlying assumptions, characteristics, limitations, and comparability of the approaches used to quantify mortality from AMR bacterial infections. We show that the global burdens of AMR estimated in previous studies are not comparable because of their different methodological approaches, assumptions, and data used to generate the estimates. The analytical frameworks from previous studies are inadequate, and we conclude that a new approach to the estimation of deaths caused by AMR infection is needed. The innovation of a new approach will require the development of mechanisms to systematically collect a clinical dataset of substantial breadth and quality to support the accurate assessment of burden, combined with decision-making and resource allocation for interventions against AMR. We define key actions required and call for innovative thinking and solutions to address these problems.

Introduction

Antimicrobial resistance (AMR) occurs when microorganisms change in ways that render the drugs used to treat the infections they cause ineffective.¹ Estimating premature mortality and the burden of disease due to AMR is crucial, both to decide on resource allocation for interventions against AMR^2, 3 and to inform the implementation of action plans at global and national levels.² With robust methods and reliable estimates, individual countries could track trends, determine the effect of actions on AMR, and compare these statistics with those from other countries. It is also crucial for policy makers to be able to compare the effect of AMR infections with other major communicable diseases such as HIV/AIDS, malaria, and tuberculosis, as well as non-communicable diseases with a large global effect, including heart disease and cancer.

The Review on AMR by Jim O'Neill⁴ estimated that 700 000 deaths each year globally might be due to AMR bacterial infections, including multidrug-resistant and extensively drug-resistant tuberculosis. The Global Burden of Disease (GBD) 2016 study⁵estimated that 126 000 deaths are due to multidrug-resistant and extensively drug-resistant tuberculosis in 2016, but the number of deaths due to other drug-resistant bacterial infections, malaria, and HIV were not estimated separately. National estimates of mortality from AMR bacterial infections have been published for places such as the USA,⁶ Europe,^7, 8 and Thailand.^9, 10 A direct comparison of these estimates is not possible because each used different approaches and data sources, including which types of infections were considered when preparing the estimates.

Despite the importance of AMR as a public health threat, the scarcity of a robust and accepted approach to assess its burden is widely acknowledged.^{2, 11, 12, 13, 14} The burden of AMR can be measured using many parameters, including mortality, morbidity, economic cost, and resource use.^1, 12 In this Personal View, we limit discussion to mortality from AMR, drawing on a combination of published evidence and expert opinion. We compare and discuss general underlying assumptions, characteristics, limitations, and comparability of the approaches that have been used to quantify mortality from AMR bacterial infections in a country or globally.^{4, 5, 6, 7, 8, 9, 10} We also propose general guiding principles and potential approaches for improving these estimates in the future. We focused on the approaches used by a review by O'Neill⁴ and the GBD study by GBD 2016 Causes of Death Collaborators⁵ because of the availability of their estimates of global mortality from AMR and their high effect on national and international stakeholders.

What is the cause of death?

Determining the cause of death can be complicated. Patients often die from a combination of underlying conditions, comorbidities, and acute complications such as a drug-resistant infection that ultimately results in death.^15, 16 It is often hard to decipher which of these conditions initiated the series of events that resulted in death. For example, did the patient die because of a drug-resistant infection or did the patient die while being infected by a drug-resistant infection? The international classification of diseases (ICD)¹⁷ is the global standard that is widely used to promote international comparability in the collection, processing, classification, and presentation of mortality statistics. Globally, health-care systems document all the patient's medical conditions using ICD codes, and certifying physicians record the sequence on the International Medical Certificate of Causes of Death. This information is then coded by a trained ICD coder to select the underlying cause of death. Information about comorbidities and the presence of sepsis^18, 19 can be determined from the combination of underlying and non-underlying (ie, immediate or intermediate) causes listed on the death certificate, which are collectively known as multiple causes of death.¹⁷ Sepsis is a life-threatening organ dysfunction caused by a dysregulated host response to infection, and the final common pathway to death from most infectious diseases.^18, 19 Nonetheless, the global standard for mortality statistics is to select only one main underlying cause of death as a single cause of death, regardless of how many conditions are reported in medical records or death certificates.

Although the ICD principle relies on the assumption that there is only one cause of death,¹⁷ the cause of death, particularly for complex illnesses, is often the interplay between two or more major morbid conditions (figure 1). Assigning a single cause of death according to the ICD principles means that most hospital-acquired infections and an unknown proportion of community-acquired bacterial infections do not feature in statistics as a cause of death because of the presence of an underlying condition that led to the original hospital admission.