The Miami COVID Project:

Visualizing COVID-19 in Miami-Dade County

Roy Williams, Anny Rodriguez, Mary Jo Trepka, Zoran Bursac, and Gabriel J. Odom

20 October 2021, FIU-RCMI

Outline

  • How the Project Began and COVID-19 timeline
  • Consulting Question: Are we better now, or worse?
  • Consulting Question: Are we reporting COVID-19 deaths correctly?
  • Lessons Learned and Challenges Overcome
  • Summary and Closing Remarks


Slides available at https://rpubs.com/gabrielodom/Visualizing-COVID19 DISCLAIMER: All data is from the Florida Department of Health unless otherwise stated. The authors gratefully acknowledge use of the services and facilities supported in part by the National Institute On Minority Health and Health Disparities of the National Institutes of Health Under Award Number NIMHD (U54MD012393), Florida International University Research Center in Minority Institutions. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

How the Project Began and COVID-19 timeline

The COVID-19 Epidemic in South Florida

  • COVID-19 is a highly contagious respiratory infection caused by the novel coronavirus SARS-CoV-2
  • The incubation period is 2-14 days, so observed data lags community transmission 1-2 weeks
  • The South Florida (SFL) metropolitan region is comprised of Miami-Dade, Broward, and Palm Beach counties
  • Roughly 6.4 million people live in these three counties (30% of the state’s total population)
  • Since March 5, 2020, Florida has had 3.63 million COVID-19 cases (1.25 million in SFL) and over 58,000 deaths (17526 in SFL)

COVID-19 Epidemic Timeline in South Florida

  • First Wave: Early March (second week) to mid-May 2020
  • Second Wave: Mid-June through August 2020
  • Third Wave: Late October 2020 through 2021 “Spring Break” Season and ending in May 2021
  • Fourth Wave: Mid-July through September 2021 (shown later)

Context: The First Wave (Mar-May 2020)

  • March 16, 2020: confirmed “Patient Zero” in Miami-Dade
  • March 28, 2020: Gov. DeSantis issues a “Safer at Home” order (a lockdown) effective April 1st and required all hospitals to report their COVID-19 data to the state daily
  • Early April 2020: representatives for local elected officials reached out to us for help understanding trends in the data they were receiving daily (example report excerpt below)

The Miami COVID Project Begins

  • In mid-April 2020, the White House released reopening guidelines based on 14-day trends in testing, positivity rates, and hospital capacity
  • Meetings with Baptist Hospital Center for Advanced Analytics team
  • Determined need for trend visualization and began analyses and website
  • Joined weekly “Miami Phased Comeback Meetings”

“Are we better now, or worse?”

“Nothing could be older than the daily news, nothing deader than yesterday’s newspaper.” - Edward Abbey

Context: The Second Wave (April-May 2020)

  • The 14-day trend of case counts and test positivity rates became negative in early May, and hospitals had sufficient capacity
  • On May 18, 2020, Miami-Dade County lifted the lockdown restrictions
  • At the end of the spring school semesters and after the Memorial Day holiday (May 25), cases began to rise

Context: The Second Wave (June-Sept 2020)

  • The Florida Department of Health (FLDoH) reported testing data for the most recent 14 days only (below)
  • Local elected and administrative officials struggled to interpret this data, lacking context necessary to compare to the First Wave

Figure: Contextualizing the Current Spread

Decision to Require Masks in Public

  • Using our figures, local officials were better able to contextualize the current state of affairs in the county than they had been using the tables/figures reported by the state Department of Health
  • A “Mask in Public” ordinance had been in place since the end of May, but it was not enforced
  • On July 2, 2020, city officials began enforcement of that ordinance
  • Note 1: Because of the viral incubation period, “current” testing data and hospitalization data was already too old to measure the current state of community transmission

“Are we reporting COVID-19 deaths correctly?”

https://slate.com/technology/2020/06/covid-death-rate-declining-explained.html

Context: Peak of the 2nd Wave (July 2020)

  • During the rapid rise of COVID-19 cases and hospitalizations, some of the officials we met with noticed that the number of COVID-19 deaths reported by the state was decreasing, while the number of patients reported to be on ventilators in the hospital was increasing
  • Based on the medical experience at the time, we expected 80% of ventilated patients to die


Was this because COVID-19 was no longer a threat?

NO!

Interpretation

  • The grey columns are the reported COVID-19 deaths for March-June, 2020, as reported on July 1, 2020
  • The other stacked columns show how the counts for the exact same days will increase over time as more deaths are reported and certified
  • Using this figure, we convinced the local officials that daily death counts were completely inappropriate data to use in making health policy decisions
  • The local offices of the health department informed us that that medical examiners offices were hopelessly backlogged (Florida required at that time for all COVID-19 deaths to be certified by a medical examiner; this policy changed in August)

Making Inference

  • We no longer have 2-3 month delays in death reporting; this “lag” is 1-3 weeks now, and we have a better understanding of the implications of this delay
  • However, even in 2021, these same shoddy and misinterpreted data are still being used to erroneously claim that the Delta variant “isn’t that bad”

Lessons Learned and Challenges Overcome

Challenges–Data Collection

  • Data dumps
  • No data at all
  • Data quality and reliability
  • No data dictionaries
  • Data in different formats
  • Vaccine Data:
    • Some municipalities in MDC over 100%
    • Higher rates in MDC than other counties or state

Challenges–Data Reporting

  • Reporting of different numbers from different agencies
  • National news reporting sensationalizing numbers from unknown sources
  • Differing metric definitions, numerators, and denominators
  • Lag in death certification and timely counts
  • Three weekly meetings at peak times

What We Learned

  • Real data is messy
  • Poorly displayed data can make it difficult to make informed decisions
  • We were blessed to have local officials who wanted to make data-driven decisions
  • We were lucky to be able to work with an incredible interdisciplinary team of experts
  • “Hindsight is 20/20”

Current Efforts

  • Our paper is here (it’s open access)
  • The emergency order giving us access to the data expired in June 2021, so we have not had access to state data
  • We have access to similar data from the CDC, but they use a different data dictionary
  • Miami-Dade County Mayor Levine-Cava required all county hospitals to report their data on COVID-19 admissions in response to the Fourth Wave
  • We have partially reverse-engineered the state health department data from the CDC data in order to have an “apples to apples” comparison across the entire pandemic

Summary and Closing Remarks

Our Prediction in July

Current State

Current State

Our Team

Acknowledgements

  • Florida Department of Health: Drs. Villalta and Suarez
  • Baptist Health: Drs. Veledar and Saxena
  • Jackson Health: Carlos Migoya and Drs. Abbo and Paige
  • FIU College of Medicine: Dr. Marty
  • Miami-Dade County: Mayor Daniella Levine-Cava, Frank Rollason, and the Office of Emergency Management
  • City of Miami: Mayor Francis Suarez, Dr. Robert Hevia, and the Office of Emergency Management
  • City of Miami Beach: Mayor Dan Gelber and the Division of Emergency Management
  • Miami-Dade League of Cities: Mayor Joseph Corradino

Final Thoughts


Please get vaccinated!


Slides available at https://rpubs.com/gabrielodom/Visualizing-COVID19


GitHub repository for this presentation: https://github.com/gabrielodom/JSM_2021_COVID_Visuals