Introduction to the guide
Coronavirus disease 2019 (COVID-19) has a wide range of documented effects. It directly causes death and disability for some people infected. However, disruption to essential health services, resources allocated to mitigation and therefore away from essential health service delivery, and the overall impact on the economy and society must also be considered within the response to COVID-19. Understanding the magnitude of all of these effects is an essential part of developing mitigation polices.
Several epidemiological models have been created to assess the potential impact of disruptions to essential health services caused by COVID-19 on morbidity and mortality from conditions other than COVID-19 illness. This guide presents models that have been used to assess these indirect impacts. The effects have been studied in various settings, using a variety of models.
The guide is intended for people who need to understand what the models say, their construction and their underlying assumptions, or need to use models and their outcomes for planning and programme development and to support policy decisions for a country or region.
Of course, an overview of models on COVID-19 is a moving target. Modellers create new models and they revise and improve established ones. Since the field is rapidly developing, it is important to note that modellers may have to overcome limitations or concerns that may be voiced here about approaches. Thus, the document may be revised to reflect these changes if such changes occur.
This document provides an overview and description of models from a technical point of view. The focus is on what the various models do, how they do what they do and the underlying assumptions on which the models are based. The document includes modules on modelling the disruptions caused by COVID-19 to the essential health services of specific health areas or conditions.
Each module will follow the same structure as closely as possible for consistency:
introduction to modelling for COVID-19 service disruptions
service disruptions in the context of the health area or condition of interest
models used in the disease of interest, their strengths and weaknesses, and their interpretation
outcomes and use of modelling studies to date
conclusion and recommendations for the use of models in the response to COVID-19.
This version of the guide (version 1) includes only the module on sexual, reproductive, maternal, newborn, child and adolescent health, and nutrition. The next version of the guide will contain more modules on other health areas or conditions.
The annexes to the guide comprise a discussion of data sources in general and an annex for each model presented, which gives details of the particular model for modellers and statisticians who wish to have this information.
Module 1: Understanding modelling approaches for sexual, reproductive, maternal, newborn, child and adolescent health, and nutrition
Introduction to Module 1
COVID-19 has a wide range of effects. It causes death and illness, but its indirect negative effects are just as important, such as the disruption to health and society, the resources that have had to be allocated to its mitigation and away from other areas, and the overall impact on the economy and society. Understanding these effects and how policies can eliminate, reduce or mitigate them is crucial.
This module focuses on the effects on health services and societal mechanisms related to maternal and child welfare and survival, as well as sexual, reproductive and adolescent health and nutrition. The effects of COVID-19 have been studied in various settings, using a variety of models. This module provides a technical overview and description of such models. The focus is on what the various models do, how they do what they do and the underlying assumptions on which they are based.
The modelling of the impact of COVID-19 on health outcomes reviewed here has four main processes:
the dynamics of the disease
the disruption caused
In principle, these represent four black boxes, with pathways between them. Models differ in the extent to which they let one or more of the boxes remain black or they open the elements of those boxes, and how they portray the pathways.
For example, the Lives Saved Tool (LiST) was first used for modelling the health impact of health service disruptions related to COVID-19. The model assumed specified levels of disruption expressed as a decrease in coverage of health interventions, and used established knowledge of the effects of interventions on health outcomes. Neither the dynamics of the disease itself nor the actual disruption level experienced in countries was modelled. This is considered a hypothetical scenario. On the other hand, there is the more integrated model of the Institute of Health Metrics and Evaluation. This model combines disease dynamics, population change, economic dynamics and effects, disruptions and health outcomes in an attempt to mimic as closely as possible realistic disruption data and interactions between multiple competing considerations of policies.
We can distinguish between the following types of models.
Models that try to estimate health outcomes depending on levels of disruption. Such models may assume likely levels of disruption, or base disruption levels on data from particular countries or regions. Other implementations of the same models, as the epidemic develops, may use disruption data from countries to predict specific outcomes.
Models that link a specific type of health intervention with disease dynamics (risk–benefit models). This category includes models that estimate the balance between lives saved by enabling children with a critical disease or injury to seek health care, and COVID-19-related deaths that may occur as a consequence of the care seeking.
Models that estimate the consequences on other diseases of disruptions caused by COVID-19. Here, a main focus is models that try to understand how HIV prevalence may be associated with disruptions caused by COVID-19.
Models that integrate disease dynamics, disruptions and outcomes. These models may limit their focus on a narrow set of outcomes or they may try to construct a comprehensive picture of the health and societal effects of COVID-19.
The purpose of the models may vary, from advocacy linked to what might happen given realistic levels of disruption, to the prediction of what will happen given what is known about the disease and development of disruptions. Models may also focus on the costs and benefits of possible mitigation strategies or policies, measured by different metrics such as lives saved or lost, or economic costs.
The above list of models does not include models of the COVID-19 pandemic dynamics. Several reviews of such models are available. A good source is the multi-model comparison collaboration that has produced both a policy and technical report (1,2). Comparisons of models of the pandemic itself are also available in the literature (3–5).
The target audience for nearly all of the models is public health professionals engaged in COVID-19 management and mitigation from a technical or policy point of view. The model of the Institute for Disease Modelling aims to provide health professionals and families with an individual-level tool to assess the benefits and risks of various care-seeking activities.