Toolkit in real life

These are examples of the Local health data analysis and communication toolkit approaches being used in real life.

I publish these about every 3 months.  Want them sent to your inbox?  Sign up here!


Building out context

Posted 3/26/2026


The situation

A few years ago, I was looking at mortality data for a county with a large older population.  I told a colleague that the county’s death rate was about the same as the state average. 

They said that was great news — because the county is older, they’d expect a higher death rate, so this must mean the population is healthier.

I then felt awkward when I had to backtrack and explain that the statement already took age into account.  


What analysts often get wrong

Analysts are trained to account for confounders in our analysis, which is what age-adjusting does.  It's a good practice because age is a strong predictor of many health outcomes.  

At the same time, for many audiences, leading with an adjusted figure can skip a step and create confusion about what the numbers actually mean.  

That confusion can create distance between analysts and their communities.  I've had this happen many times in my career.


What to do instead

The "Producing meaningful reports" section of the toolkit, starting on page 6, suggests that analysts can:

Start with the count (number of deaths) to ground the conversation:

Then start building out context for what that number means. First, we can calculate a crude rate to help folks understand how much the health issue is impacting our population:

Now we can build out context a step further by comparing ourselves to a benchmark.  I like to do this for both crude and age-adjusted rates. In this example, I found that our county's crude death rate was 50% higher than the state average, but after adjusting for age, the rates were almost identical. Here's how I explained this:


Why it matters

Starting with something concrete helps people follow what's happening, which reduces misinterpretation and builds trust in both the data and the analyst.

In this situation, this approach was very well-received and helped build a trusting and mutually respectful relationship with my colleague.  In my experience, this step-by-step approach helps people understand what they’re seeing without getting lost.

As epidemiologists, we often assume the issue is data literacy. In many cases, it’s not. People can follow the logic - we just need to do a better job explaining it, starting with meeting them where they are.