Florida COVID-19 trend analysis: Hospitalizations Doubling Every Five Days

by | Apr 2, 2020

With the evening update now posted by the Florida Department of Health, here’s the latest run-down of where things are headed and what to look for in the next week:


Daily hospitalizations are the number that really matters, much more so than the total number of cases. As I’ve explained before, it’s these serious cases requiring hospitalization that are driving the policy decisions affecting our economy. Lives are at stake – a record 43 deaths today in Florida were attributed to coronavirus, and within weeks, experts predict we’re going to start approaching the limits of Florida’s ability to care for patients with the virus. But those same experts projected double the hospitalizations we currently have.

Note: We have asked FDOH for discharge numbers and total recovery numbers. We were informed they are working on getting that data for us.

According to a new dashboard released by Florida’s Agency for Health Care Administration, Florida currently has exactly 23,898 23,980 (as of 10am Friday) available hospital beds – almost 3,000 more than what some widely-cited models project. But we’re starting to fill them at a steady clip. Note the slow, upward curve of the red hospitalization line, a clear indication we’re still seeing some level of exponential growth (use pinch to zoom on mobile):

Take a closer look at the hospitalization numbers on the red line. Starting back on March 19th, it took 4 days before the total doubled. The doubling interval held steady each day until March 23rd, when it starting doubling just a little faster, an interval just slightly longer than every three days. That trend continued through March 25th, when it abruptly slowed to an interval of five days, where it currently stands. In fact, today was the first doubling we’ve had since March 27th.


The most widely cited model for what we can expect in terms of hospitalization rates over the coming weeks is from the Institute for Health Metrics and Evaluation (IHME). If you clicked the link, you’ll still need to select Florida in the drop down tab to see what I’m about to show you.

IHME is currently predicting we will hit our peak hospital resource usage on May 3rd, exactly one month from now. Their model predicts we’ll need about 16,963 hospital beds that day to treat all the seriously ill COVID-19 patients. As I mentioned above, Florida currently has about 22,000 beds available, giving us a buffer of about five thousand beds (which is also 2,000 more than what IMHE has in its model for Florida).

If you think that’s good news, you’re mistaken. If we have 16,000 seriously ill coronavirus patients in hospitals across the state, our health care system and our doctors, nurses and other medical personnel are about to be pushed to the limit. This is why Governor Ron DeSantis finally relented and ordered a statewide lockdown, to try to flatten this very curve. Note that dark purple dash in the chart. That’s the projected trendline. The solid dark purple line is the total beds available (again, this chart shows 2,000 lower than Florida actually has – good news for us). But the lighter purple shaded area is the margin of error. Things could turn out much better…or much worse:


Don’t worry. I found some. If we slide the graph above back to today’s date, we see that IMHE’s hospital resource graph hasn’t been updated in a couple of days, and its assumptions predicted we’d already need 2,612 hospital beds today, April 2nd, and three hundred more by tomorrow (see text box in the lower left):

As mentioned above, we only added 177 new hospitalizations today, and we’re not likely to double that in a single day tomorrow. Even if we did admit 300 new patients to the hospital by tomorrow, our total beds in use would still only be 1,400 patients vs. IHME’s projection of 2,900+. Let’s hope the trend continues and we only see half the hospitalizations they predict.

In less important news…


Despite other media outlets reporting the jump in cases as if it’s another harbinger of doom, there is nothing surprising about today’s total number of new coronavirus cases. The more we test, the more cases we find. Today we nearly doubled the number of tests performed yesterday. That’s a 100 percent increase in testing, while new cases only went up by about 20 percent:

Obviously we want to get to the point where the percentage of positive tests starts to trending toward zero. We’re not there yet. Here’s a look at positive testing percentages. Note the steadily climbing cumulative positive rate (red line). Though today’s big drop off (blue line) pushed the red line a bit lower, we need them both going down, down, down:

By now, it should be clear to all that we’re in this for the long haul. IHME’s project has the peak hitting Florida in early May, and the last case leaving Florida in July.

We’ll continue to report on the trends we’re seeing in the data as new trends develop.


I’ll leave you with an explanation about the data we’re using – a number of people have noted that it differs slightly from the data they have personally collected or the data posted by FDOH. For the record, all of our data is directly from FDOH and Florida Department of Emergency Management (FDEM) sources. We’ve been tracking this data since March 1st when the first COVID-19 cases were identified. However, over the past month, FDOH has revised their data reporting methodology several different times and ways, which accounts for some of the differences. We have made efforts to focus only on specific datasets so that we can report as consistently as possible.

Those data sets are:

  • Total cases reported in Florida, regardless of residency status
  • Total new cases, regardless of residency status
  • Total tests and daily tests administered
  • Total and daily positive and negative tests
  • Total hospital beds in use
  • Total hospital beds in use by COVID-19 patients
  • Daily new COVID-19 hospital admissions
  • Total deaths / daily deaths
  • Total recovered (this data is still not available)

The source for all Florida-specific data used by The Capitolist from March 19th forward is pulled from floridadisaster.org/covid19/covid-19-data-reports/. Prior to March 19th we collected all data from the FDOH COVID-19 dashboard.

There have been significant inconsistencies in the data reported by FDOH, and some of their reporting changes make it difficult to find the correct data. We work to mitigate those reporting inconsistencies wherever possible.

The problem is that some of the classification decisions FDOH and DEM made early on with reporting have served to make the entire data reporting process more confusing and probably more cumbersome for everyone, including other media outlets and the state’s own staff.

Chief among the complaints is the inexplicable decision to report separate “Florida residents” from “non-Florida cases” who test positive and/or receive treatment in Florida.

Does anyone care which state issued a COVID-19 patient’s driver’s license? If they’re in Florida, they’re a concern for everyone. Most of the folks I speak with just want to know how many people in Florida are testing positive for the virus. Nobody wants to know where they legally reside.

Non-Florida residents testing positive in the state now comprise a sizeable chunk of the total COVID-19 population (more than 300 cases). That’s enough that any omission of those cases from our dataset throws off some of the calculations by a noticeable margin. Wherever possible, we try to use total case data, regardless of legal residency in Florida, but any discrepancies between our data and that reported on “official” dashboards can primarily be explained by this single difference. Worse, FDOH went through several days where they were reporting all cases, then switched to reporting only Florida cases, and finally switched back to providing data for both types of cases – and this all happened at different points in time in different data modules, too.

Because of the lack of certain information, we are forced in many cases to compute our own data points from what is provided by FDOH. For example, FDOH has its own methodology for how it reports the total number of daily tests performed, which involves attempting to remove duplicate test results and other information. For our purposes, we simply compute daily tests by first tabulating “total tests” from all positive cases and all negative cases before subtracting the previous day’s total. We also attempt to ignore invalid tests when those numbers are reported and known.

Hope that helps better understand how we produce the charts we use in our stories.

Stay safe and minimize contact with others as much as possible. More soon…







  1. CP

    Please start looking at Florida and the virus in regions: south FL (HOT), Central FL (medium), Northeast FL (low to medium), Northwest FL (low to medium).

    The south of the state is really in need of help. If the people in the rest of the state keep their numbers low, they can help the south. The issue isn’t total beds – it’s actually answering this question:

    When will south Florida run out of capacity? That’s going to be a major problem.

  2. Robert Godwin

    Best information out there. thank you! And, you are polite about the incompetence of the state, but perhaps now isn’t the time for grace toward our government.

    The state doesn’t know the discharge and recovery numbers? I’m speechless. So, that means the state can’t know the number of active cases? Let’s work through the emotions together: Disbelief, frustration, anger, despair and then anger again.

    Let’s imagine in the next week, we have numbers like NY, and the state of Florida couldn’t tell residents which hospitals have available beds, because they are not actively tracking hospital discharges. On the John Hopkins website, for Florida, there are no recovery numbers, because our state doesn’t track the number of recoveries.

    Apparently, keeping the golf courses open was a bad idea, because all of our municipal employees are practicing their short game, instead of following up with primary care physicians to track recoveries.

%d bloggers like this: