Analyzing Covid-19 hospitalization data from the Department of Health and Human Services, we found which hospitals filled their intensive care units to capacity each week throughout the pandemic. Using that analysis, we created a list of the most stressed hospitals in the country and identified 20 hospitals that had been at or above capacity for more than a year.
We then spoke to doctors and nurses from those hospitals to hear what they’re experiencing and contextualized it with data and visuals that let the reader see which states and which hospitals never caught a break.
We illustrated the stress that health care professionals, especially those in overburdened hospitals were going through. While waves would come and go during the pandemic, these hospitals had no real reprieve and have been at near-constant capacity since the pandemic hit.
The story was published at a time when vaccine hesitancy was at a high and public sector unions were actively protesting vaccine mandates. The story illustrated the relationship between low-vaccinated areas and overpacked ICUs.
Using Python notebooks, we pulled hospitalization data from Health and Human Services and took the weekly average of staffed ICU bed occupancy and divided it by the weekly average of staffed ICU beds. That formula gave us a hospital-level utilization rate for reporting hospitals.
HHS did not include hospitals with fewer than 4 patients hospitalized in a reporting period that were not included in the data. We then aggregated the utilizations to county, then state level, which informed our reporting, but also was used in our visualizations.
State-level results were broken up by reporting week, then visualized as graphics using D3.
County-level results were aggregated by month, grouped by FIPS code, joined to shapefiles, then visualized on multiple D3 maps to illustrate the maxed out regions by season. Readers were able to see that even when waves came and went, some parts of the U.S. had consistently maxed ICUs.
What was the hardest part of this project?
One of the hardest things about this project was understanding the data. We talked to multiple experts to establish what is normal and what is not normal. Intensive care utilization is a bit abstract to the normal person and hearing from doctors, nurses and associations about how far out of normal constant capacity helped illustrate the urgency.
We’re subjected to near constant headlines about hospital capacities during the pandemic, and it can be easy to get numb to it as waves come and go. But finding the most stressed hospitals, calling them to verify what’s going on the ground puts people to the data.
What can others learn from this project?
We talked to staffers who handle HHS data to verify any pitfalls we’d encounter with our analysis as well as verifying the accuracy of the data itself. Since the pandemic it’s easy to pull data sources and turn them into dashboards, but verifying from the source as well as relying on documentation assembled by a consortium of experts helped a great deal.
With every hospital called, we ran the data analysis by them, and asked if that sounded accurate. Many of them said “yes,” and quite a few said their current ICU numbers were a little higher than that.
The major takeaway that other journalists can learn is to verify your findings at any available avenue.