Understanding Rates
On the Congressional District Health Dashboard, we present metrics related to deaths (mortality) and births (natality) as rates. Other metrics on the Dashboard are presented as percentages or averages. In this blog, we’ll break down why certain data are most effectively communicated using rates, and how to make sense of rates. We interact with different types of rates all the time – for example, if the speed limit is 50 miles per hour, that’s a rate – of distance traveled per unit time. Or, if someone making deliveries earns $20/hour, that’s also a rate – of dollars earned per unit time. Technically, percentages are a type of rate. If 60% of people have diabetes in a district, you could communicate the same information as a rate: “There are 60 people with diabetes in this congressional district per 100 population.” When we present mortality rates on the Congressional District Health Dashboard, we use “per 100,000 population.” For example, in Ohio’s 10th congressional district in 2022, there were an estimated 7 firearm homicide deaths per 100,000 population. Why are these metrics presented as rates, when others use averages or percentages?
Averages present the typical value for an area, time, or population in a single variable, like air pollution – particulate matter or life expectancy. We could consider providing the number of deaths in a district as an average over time (e.g. per year), but rates are better at accounting for differences in population size across places. For example, what if we were interested in using data from the City Health Dashboard website to compare colorectal cancer deaths in Kettering, OH, with deaths in the surrounding congressional district, Ohio’s 10th? (Remember, on the Congressional District Health Dashboard you can identify which cities overlap with each district.)
Let’s say the average number of colorectal cancer deaths in Kettering, OH from 2019-2021 is around 10 per year, and in Ohio’s 10th congressional district it’s around 100 deaths annually. These numbers are misleading though, because Ohio’s 10th has a far larger population than Kettering (~800,000 in the district vs. ~60,000 in Kettering). We need to ‘standardize’ these numbers - adjust them so they are calculated for a common population size - to be able to determine which geography has a higher “burden” of colorectal cancer deaths.
Why not use a percentage, then? Percentages, as noted above, are a type of rate calculating an outcome per 100 of something. The percentage of colorectal cancer deaths in Kettering would be ~0.0122%, and in Ohio’s 10th would be ~0.0128%. Using percentages, we can see that deaths from colorectal cancer are very similar between the city and its district in 2021. But it’s hard to interpret and compare such small-decimal numbers. And presenting percentages this small runs the risk of understating the importance of different public health problems.
When calculating metrics for rarer outcomes (for example, colorectal cancer deaths are much less common than cases of diabetes), we strive to present numbers that clearly and accurately communicate the burden. For instance, on the City Health Dashboard you can find the number 12.2 colorectal cancer deaths per 100,000 in Kettering, and on the Congressional District Health Dashboard, 12.8 per 100,000 in Ohio’s 10th, indicating that Kettering has a slightly lower colorectal cancer death rate than Ohio’s 10th. Presenting our colorectal cancer death metric as a rate provides comparable, easy to understand numbers. You could interpret these numbers as: “If 100,000 people lived in Ohio’s 10th in 2021, then 12.8 people would have died of colorectal cancer.”
(Important note: These are simplified example calculations that don’t incorporate the age-adjustment approaches also used in mortality metrics. We do not recommend that users try to reproduce these calculations. Please see the Dashboard technical document for more information.)
How do we know if 12.8 colorectal cancer deaths per 100,000 in Ohio’s 10th is high, low, or average? This is where the Dashboard’s scalebars come in handy.
In this screenshot, the smallest number on the left side of the scalebar (8) represents the country’s lowest district-level colorectal cancer death rate in 2021, and the largest number on the right (26) represents the highest. Ohio’s 10th is on the lower end of the scale bar, but it’s near the national average of 14.1 deaths per 100,000 - denoted by the triangle at the bottom of the scale bar labeled ‘United States Average’. This tells us that Ohio 10th’s mortality rate is just below average.
The Dashboard’s mortality data can be useful for highlighting disease burden and disparities. When sharing these rates, we recommend the following strategies to help people understand them better:
Include a brief explanation of rates and “per 100,000 population”- feel free to use this blog as a source!
Compare rates between groups or districts/cities to help people understand if rates are low, high, or similar to the average rates.
Now that you’ve learned why mortality metrics are presented as rates, and how to interpret these data, we hope that you feel more resourced to utilize mortality data in your work.
