Data Snapshot: Flu makes a late stand in some parts of the US.
Michigan shows a late surge. Other regions have seen improvements.
Overall, hospitalizations of all three major respiratory viruses we track (Covid-19, influenza, and RSV) have decreased markedly from their late-December/early-January peaks this season. (Explore the data on the Inside Medicine dashboard.)
Now you may recall that I recently wrote that, anecdotally, I felt that I was seeing an increase in flu-associated hospitalizations in the Northeast.
Was I right? Or was what I was seeing due to random chance?
Updated data now show that my observations were in line with regional trends. The closest state to me that the CDC tracks is Connecticut. Things bottomed out in Connecticut in late January, but had increased by 77% by late February, which is when I posted my frontline observations about a possible late uptick in flu.
While things have improved on the flu front overall nationwide, there are indeed a few regions where influenza has been on the rise as of late, with late season peaks.
So, for this week’s Data Snapshot, here are flu-related hospitalizations (per 100,000 people) for the 14 states that the CDC reports on each week, from the apparent peak of the season for most regions this year (late December/early January), through the end of February.
As you can see, some places like Georgia and North Carolina are way better off than their flu peaks. Conversely, in Michigan, by late February things were getting worse, reaching new highs for the year. Data from this past week actually reveal that flu hospitalizations in Michigan continued to rise in early March, though I did not include early March in this week’s graph in order to make the larger points here. (You can explore all of the most recent data here on an interactive graph I made from our dashboard.)
Checking my work.
In another recent post, I wrote about how sometimes clinicians like me can’t actually tell what’s happening on the frontline, because our sample sizes are too small to observe what is much more clear in big data. It is easy for us to mix up noise and randomness with a real signal that is statistically meaningful. Yes, our eyes can deceive us. But I’m always interested in whether frontline observations match big data in a measurable way.
So, I ran some math.
Based on the Connecticut data, my hospital’s size, and how often I work, I wanted to know if I even had the ability to notice the mid-to-late February increase in influenza in the Northeast during my ER shifts. (Remember, even if a disease rate increases 100-fold from 1 in a billion to 1 in 10 million, I’d probably not be able to notice that in a week, month, or even year’s time, despite the two orders of magnitude difference.)
Long story short: Yes. The increase in flu rates in the Northeast that occurred in February basically amounted to me seeing less than one flu case per shift in late January to 1.7 cases per shift a month later. Over the course of 10 shifts, that would amount to around 9.5 flu admissions versus 17 admissions. I think that’s enough of an increase to be noticeable to a working clinician like me. So, in this case, my eyes had not deceived me.
Still, I’m glad I checked. I always want to be accountable to you when I am linking my frontline experiences with the latest high-quality data.
Thanks as always to Benjy Renton for maintaining our fantastic interactive dashboard!
Comments? Questions? Ideas? Please chime in with your thoughts!