The Epic EHR has a tool called Slicer/Dicer that allows clinician-users to set up qualitative analyses of our populations in sophisticated ways. Of course this doesn’t replace the need for report-writers and more sophisticated analyses. But it is amazing what an informaticist can come up with, sitting in an Incident Command Center on an Easter Sunday with unusually few escalation phone calls to deal with.
For example, the curve above shows Influenza Positive test at UCHealth (12 hospitals, 600 clinics) over that past 3 winters: 2018-2020. Be cautious about interpreting the data: UCHealth has grown in number of clinicians and in patient volume, behavior of testing for “flu” may have changed. But it does look like the annual peak of flu positive patients is Jan or Feb each year.
Taking this further, our lab distinguishes Influenza A from B, and looks like “B” positive peaked in December vs “A” peaking in February.
Respiratory Syncitial Virus (RSV) peaked in February.
Rhinovirus peaked in September.
The “other” coronaviruses peaked between December and March.
Human Metapneumovirus peaked in March.
Finally, Our Coronavirus RNA test shows an ongoing increase (that last column showing Zero is an artifact of delayed reporting during my report run).
These are of course Lagging Indicators: trend lines that occur AFTER the fact: patients are in our hospitals, or are positive healthcare workers with symptoms. The constraint of insufficient testing kits to test everyone who has symptoms and indeed everyone who was exposed or has concern, gives us very little surveillance data to look forward for future outbreaks. More on surveillance ideas in an upcoming post.
It does occur to me, that in the coming months and years, that Medical Education could be turned on its head. In the past, I was clever enough to show our medical school leaders that this same Slicer tool could “make the textbook come alive.” For example, a student could create a graph, from existing UCHealth patient de-identified data, that the percent of patients with hypertension increased if you compared those with a BMI of up to 20, then 21-25, 26-30, 31-35, 35-40, and then greater than 40. You could see the that the percentage increased from 5% into the 32% range. Voila: possible relationship between Body Mass Index and prevalence of hypertension!
Repeat with diabetes, high cholesterol, asthma. See what blood pressures are typical for patients on a particular BP medication.
And for our current topic, have students figure out when respiratory viruses peak over the year, instead of reading a book chapter on ‘Pathophysiology of viruses.’ That would be a med school class I’d like to take. Maybe have students help with our CURRENT problem of trying to use our EHR to detect signal for patients about to deteriorate for Covid-19.
CMIO’s take? The EHR is becoming an integral part of how a modern doc takes the deluge of health data and uses that power for good.