Where are COVID-19 data?
Various organizations collect COVID-19 data but the main responsibility for data collection lies with governments. Most countries maintain health information systems which collect and manage data from all public health facilities and produce regular reports. Countries report mortality data and cases of specific notifiable diseases to the World Health Organization which publishes them in annual reports. National governments regularly publish weekly numbers of infections and deaths from COVID-19. WHO and other international other agencies maintain global dashboards showing the numbers of cases and deaths worldwide. Some open data sites accumulate and publish COVID-19 datasets.
WHO maintains a global COVID-19 dashboard and daily situation reports. Many other organizations publish dashboards showing COVID-19 data and provide downloadable data for further analysis. The opening data page of this website describes the principles of publishing open data.
Can we trust COVID-19 data?
COVID-19 data are everywhere but we cannot always taken them at face value. Users must understand where the data come from and what they mean to decide what conclusions they can make. This is true not just for COVID-19 but for all types of data. Scientists spend their lives looking for the best ways to collect and analyse data to answer questions of public importance. They must also explain their conclusions clearly to different audiences. But they are not always successful – and some people choose to deliberately misrepresent data.
There is much advice to help users decide whether they can trust data and statistics; one excellent brief is A dozen rules of thumb prepared for journalists by the Royal Statistical Society. We also refer readers to the data quality and information integrity page of this website.
What do the COVID-19 indicators mean?
Indicators use data to describe situations as simply as possible. COVID-19 indicators describe the dynamics of the virus in a population, for example, its rate of transmission or reproductive number; or they describe its impact on a population, for example, its incidence, or case fatality rate. Indicators either summarise a set of raw data or the estimated results from a model. To understand and trust the published value of an indicator, we need to ask: What does the indicator mean and how was it derived?
The Royal Statistics Society has published A statistician’s guide to coronavirus numbers.The health indicators page of this website describes types of indicators and the measurement of indicators page explains how they are measured.
How are COVID-19 cases counted?
International dashboards show the dynamics of the COVID-19 pandemic by recording the number of confirmed cases reported by countries. At least at the start of the pandemic, most countries were only able to report cases of infection in people who received hospital treatment where testing was available. Some countries do not have sufficient testing facilities even in hospitals. Other countries seek out and test suspected cases in the community and conduct sample surveys of the population. Clearly, the number of infections in a population will be considerably larger than the number that reach hospital. Until a country has set up a comprehensive and reliable testing system, the reported number of confirmed cases will be the most severe cases. It is difficult to compare counts of infections across geographies and time, given varying capacities to test.
The World Health Organization provides case definitions for classifying a patient as a suspected, probable or confirmed case of COVID-19 in this document: WHO Global Surveillance for human infection with coronavirus disease (COVID-19). The health indicators page of this website explains the types of indicators and the measurement of indicators page explains how to measure them.
How are COVID-19 deaths counted?
Counting deaths is not as easy as it might seem. In high-income countries, hospitals, care homes, families and funeral directors report each death and its certified cause to the civil registration office. A physician will only record COVID-19 as a primary or underlying cause of death with certainty if the patient has been tested for the virus; otherwise, the physician will interpret signs and symptoms. The registration office provides regular summaries of the numbers of deaths and their causes broken down by age, sex and location. But such systems do not exist in many low-and middle-income countries. Even when relatives report deaths, a physician may not have certified the cause. Sometimes, a health worker visits the family to ascertain the cause of death retrospectively using a technique called verbal autopsy.
The United Nations Statistics Division provides an overview of how each country’s registration system is dealing with COVID-19. Impact of COVID-19: Maintaining Civil Registration and Vital Statistics during the COVID-19 pandemic. The civil registration and vital statistics systems page of this website explains how deaths are recorded and the demography page describes verbal autopsy.
Why do scientists build COVID models?
Most governments listen to scientists when deciding how to control the spread of the virus. Scientists build models that simulate what could happen given certain assumptions about how the virus will behave if the government implements specific strategies. For example, at the start of the outbreak, the UK Government relied on predictions of the number of hospital beds that they would need to provide using results from a model that Imperial College had built. Different models will produce different results. The differences depend on the assumptions made by the modellers, the type of model they have built and the data they have input into the model. Modellers must present their methods and data for review by their peers who ascertain if the predictions are valid. They must also explain to users the circumstances in which the predictions apply and the uncertainty around their predictions.
This blog by PATH provides a clear overview of modelling for COVID-19: What is “the curve”? Making sense of COVID-19 models. The statistical estimation page of this website describes different types of models and how they are used to make estimates of global health indicators. The spatial and spatio-temporal modeling page explains models that predict events in both time and space.