Overview of GBD Dimensions

Kara Estep, Director of Research Operations at IHME, presents an overview of the dimensions of the Global Burden of Disease (GBD) study.

Video transcript

This transcript has been lightly edited for clarity

Today I am going to provide a high-level overview of the different dimensions of GBD.

I will walk through the different components of GBD shown here.

The GBD includes estimates of the burden due to numerous diseases, injuries, and risk factors, with these estimates being produced across many locations, years, ages, and for males and females.

First, let’s look at causes in GBD.

We make estimates for more than 370 diseases and injuries and organize them into three categories: communicable, maternal, neonatal, and nutritional diseases; noncommunicable diseases; and injuries.

These are organized into what is called a cause hierarchy.

At Level 1, there are just these three cause categories.

As you move down the hierarchy, each category becomes more detailed.

Level 2 are the Level 1 categories broken down into more specific disease and injury groups.

We’ll use enteric infections as an example of a more detailed category of causes below communicable, maternal, neonatal, and nutritional. Enteric infections is a Level 2 cause grouping, and below that we have “typhoid and paratyphoid” as a Level 3 cause grouping.

Finally, below that is the most detailed cause level, with typhoid as an example of a Level 4 cause.

This hierarchy is mutually exclusive – there’s no overlap between causes at each level.

This hierarchy is also collectively exhaustive. At each level, it captures all causes of mortality and morbidity.

So, if you were to add up the number of deaths due to the two Level 4 causes, typhoid and paratyphoid, that is the total number of deaths for the Level 3 cause grouping of typhoid and paratyphoid combined.

These levels represent increasing levels of detail, from broad categories to specific diseases and injuries.

These figures can give you a sense of the different level of detail produced.

This is a figure that we call a treemap, which is a rectangular pie chart. This one is showing the number of DALYs by each set of causes, within each level.

You can see Level 1, the highest level at the top left, all the way down to the most detailed, Level 4, on the bottom right.

Like the cause hierarchy, the risk factor hierarchy organizes risk factors into levels.

The three Level 1 risk factors are “behavioral,” “environmental and occupational,” and “metabolic.” And all more detailed risk factors fit within one of these three categories.

In some ways, the risk factor hierarchy is similar to the cause hierarchy, with each level more detailed than the last.

In this example, we can see the Level 1 environmental risk factor, and below it the Level 2 risk factor of air pollution. Below that is the Level 3 risk factor of particulate matter pollution. And this Level 3 is then broken down into two Level 4 risk factors – ambient particulate matter pollution and household air pollution.

Note that not all Level 3s have a Level 4 beneath them. In this example, you can see ambient ozone pollution and nitrogen dioxide pollution are the most detailed level we estimate.

Unlike the cause hierarchy, the risk factors hierarchy is not collectively exhaustive. In other words, not all burden can be attributed to a risk factor.

In GBD, we produce estimates for a standard set of locations.

We use a location hierarchy starting with Global, then seven super-regions – these are shown as similar colors on the map (e.g., the sub-Saharan Africa super-region is shown in shades of red).

Then 21 regions – these are shown as different colors on the map (for example, each region in sub-Saharan Africa being shown in a different shade of red).

And in the next level below are the 204 countries and territories that are estimated.

Finally, we also include subnational estimation, for example at the state or provincial level, for some countries.

We produce estimates for any country with a population greater than 200 million, as well as some additional countries.

Collaborations with in-country partners and data availability are key factors in determining where subnational analysis is included.

In most rounds of GBD, we have added new subnational analysis for a small number of countries and plan to continue to do so.

With every new round of the GBD, the entire time series is re-estimated from 1990.

In other words, GBD 2023 includes revised estimates for the year 1990 as well as the most recent years included, 2022 and 2023. This allows estimates to be comparable to each other by using the same methods and approaches for all years. We are also always acquiring new data sources for many years in the past, which help improve our estimates.

Some indicators go further back in time, such as cause-specific mortality, which is estimated back to 1980, but all GBD results are produced for at least as far back as 1990.

We make estimates for five-year age bands from age 5 to 95+.

For ages 0 to 5 years, we have more detailed age groups – the first week of life, then the first month, then 1–5 months, and so on, since health patterns change rapidly in this age span.

In addition, we make estimates for a selection of aggregate age groups, by adding together the detailed results, for example, estimates of mortality for all age groups under 5 combined. And of course, estimates are available for all ages combined and age-standardized estimates.

We make estimates for male, female, and all sexes combined.

I’ll note that current data quality and availability only allow the GBD to estimate by sex for males and females, and does not allow estimation by gender, gender identity, or intersex populations.

All results are available publicly on our data visualization tool GBD Compare and the GBD Results tool. These tools allow you to explore visualizations of the data and to download the results.

These tools include more than 555 billion estimates of diseases, injuries, and risk factors and include estimates of mortality and morbidity for each – as well as additional indicators such as population, life expectancy, fertility rates, and more.

These results were estimated using more than 320,000 data sources.

I hope this was helpful. Thank you for taking the time to learn more about GBD.