Pandemic Parenting With Health Data

For our family, the pandemic started in earnest on March 14. We returned from a Spring Break vacation the day before and the reality of shelter in place guidelines washed over us for the first time. A routine trip to the supermarket revealed shelves emptied of canned and paper goods and long lines of quiet, uncomfortable strangers unsure of how to evaluate each other’s health. As a parent, husband, son, and healthcare data science professional, the most fascinating aspect of how our community has responded to this pandemic has revolved around the struggle of applying data to making decisions for our families.

Returning to the end of school

The most obvious and near-term decision in front of us was how our two high schoolers would return to the fourth quarter of their school year. Our school district did a good job of communicating by email, on their website, and even holding online forums. They were honest about not fully understanding all the details and moving parts, but they vowed to try – giving a worthy nod to agile experimentation, acknowledging the potential for failure, listening to open-ended feedback, and relaying their focus on taking care of the kids. The school set reasonable but firm expectations of performance, understanding the effect drastically changing educational models would have on students. Even though other countries were experiencing spreads of coronavirus, there were less than 20 cases per day being reported in Arizona at this time, the Northeast had yet to see its dramatic outbreaks, and the country was poised to respond. Online school, grocery delivery, and continued monitoring of infection data all appeared to be reasonable steps.

Executive orders and State guidelines were issued in time for Arizona to keep infections under control. Although estimates projected potential infections as high as 7,000 per day in our state, we were averaging about 300 per day through the end of the school year. A few less visits to see grandparents, a few conversations about study-from-home rules and setting up a little structure (“yes, you still have to wake up that early”), a lot of extracurricular creativity, especially considering the particularly blistering, summer heat, and things were fine – not normal and not optimal, but fine. The pandemic metrics were acceptable in Arizona well into the middle of May.

Time for a little math (sorry)

Having worked in healthcare for more than 25 years, with half of that time spent dedicated to healthcare data science, the two measurements most interesting to watch through this pandemic have been the effective reproduction rate (abbreviated as Rt) and the percentage of patients with COVID-like illness being seen by hospitals (abbreviated as CLI).

The Rt is important because it measures the rate at which the infection is spreading. An Rt of 1.0 means that for every person with a COVID infection, that person is infecting an average of one additional person (1 person with an active COVID infection divided by 1 person newly infected by the person with an active COVID infection equals 1.0). When the Rt is above 1.0, the rate of infection is growing and when the Rt is below 1.0, the rate of infection is shrinking. The higher the number, the faster the spread. Lower is better, and anything below 1.0 is good.

Measuring CLI is important because it enables us to track the demand experienced by hospitals, how many severe infections are being experienced, and to compare infection rates over time for diseases that may be hard to tell apart from one another. Because COVID symptoms can be similar to those of common colds and flus, knowing the percentage of people seeking care for COVID-like symptoms tells a story of community-wide infection trends, and understanding the direction we are heading is necessary for us to adjust our course. CLI is calculated by adding up all the patients seen in a hospital with COVID-like symptoms and dividing that number of patients by the total number of patients seen at the hospital (a hospital seeing 10 patients with COVID-like symptoms and serving 100 total patients would have a CLI of 10.0%). Lower is better here, too, and the State currently sets 5.0% as their benchmark for good.

Data for Maricopa County, Arizona through October 4, 2020 which will be refreshed on October 29, 2020, representing a delay of 25 days. https://www.azdhs.gov/preparedness/epidemiology-disease-control/infectious-disease-epidemiology/index.php#novel-coronavirus-schools

Messing up Memorial Day

By the middle of May, we all felt a little exhausted from exercising the new rules of the past three months. The data made us comfortable enough to trigger an end of school trip up North for time away from the heat and our sheltered routines. Even though Rt was around 1.2 and slightly climbing, infections were stable at less than 500 new cases per day, and CLI had been below 4.0% for several weeks. On the drive to our remote cabin, we passed by packed rows of tents in campgrounds and creeks filled with people happier to be breathing fresh air than I had ever seen. We all commented on how this appeared to be a predictable, over-reactive escape and wondered out loud if it would have an impact on the spread of the virus. Sure enough, it did.

Within two weeks, infections began to explode. CLI exceeded 15% and daily, new cases increased by more than ten times from less than 500 per day to over 5,000 per day. Shelter in place guidelines had been relaxed, people were ready to get out, and they did. People were not paying any attention to the data. Around the dinner table, we talked about the surge of illnesses. Our boys brought the remarkably mature perspective to the conversation that they understood being inconvenienced was better than being responsible for a friend or family member catching COVID. We talked about the data – how it was collected, what it measured, and what it meant. I had a proud Dad moment or two.

June was lousy as we were repressed by the summer heat and the virus. We kept talking about the metrics and the science of COVID as a family. There were awkward conversations with parents of our boys’ friends to evaluate where their families fit on the scale between cavalier and conservative as we tried to preserve a level of compromised socialization between our boys and their peers. Mask orders went into place by late June and led to a notable improvement in all the COVID-related metrics throughout July. The trend was moving strongly in the right direction, even encouraging us to take another trip up North. By late August, things were looking up. New infections were consistently hovering around 700 per day, Rt was below 1.0, and CLI was slightly above 2.0%. All reassuring news.

Confirmed COVID cases in Arizona by day from February 1, 2020 through October 24, 2020. Newly confirmed Illnesses may take as many as 7 days to be reported here. https://www.azdhs.gov/preparedness/epidemiology-disease-control/infectious-disease-epidemiology/covid-19/dashboards/index.php

A new school year

Our boys returned to school in early August. If you asked them directly, they would say they did not like it, but I believe they quietly appreciated a few small aspects like being able to wake up late (no bus to catch or hair to comb first thing in the morning) and the supportive comforts of home (lots of mid-day snacks and increased freedom of schedule). The return to school also triggered the first haircut of the pandemic not delivered by Dad in the backyard, and it was a refreshing milestone.

New district and school administrations were also refreshed, but they now communicated far less than in the previous school year. Frequent emails and regular web meetings for parents with the principal had disappeared and been replaced by just-in-time pronouncements posted to the district website. Full time, in-person teaching would not be offered, but families could choose between an all online option or a hybrid option which included two days of on-site teaching with three days of online learning. There was no clear indication of how or why the administrations arrived at their conclusions or what recommendations or data were being considered, but the options seemed reasonable, if not complex to manage. We looked at the current data and the lessons of the Memorial Day outbreak. We decided on the all online option.

Bumps and bruises aside, the return to school this year was much smoother for students than the end of the last school year. Processes were more streamlined and teachers were more prepared, but not fully. The first quarter passed by, mostly mundane. I was impressed that our boys kept up their grades and stayed as engaged as they did, even though they were clearly missing many aspects of their school community.

A bright spot that I credit for strengthening our boys’ psyches was their return to interscholastic sports. They are both swimmers and we made the decision to allow them to return to the pool. The small size of their team and the particularly low rates of infection in our community all supported our boys being able to enjoy their sport and their peers. A few hours a week with their teammates appeared to make a lot of difference to them.

Family decisions on how to deal with COVID had been easy to this point. Not sure what is happening in the beginning of a pandemic? Online school was the only path. A summer overrun with infections and restricted businesses? Find creative ways to enjoy our time together. Return to school on the backside of a massive spike of infections? In-person school is still not an option, so back to online learning. The options presented agreed with the data, or at the very least did not contradict it.

Now in the middle of October, our boys are returning from Fall Break to the beginning of the second quarter of school. Our school administrations are effectively requiring a return to in-person learning – an online learning option is technically available but it offers no honors and few elective courses, making 60% of our boys’ classes unavailable online. An improvement from the first quarter, school administrations have communicated metrics for when in-person learning would and would not be allowed going forward. Established, data-driven thresholds for when in-person, hybrid, or online learning are appropriate limit debate and make decisions easy. I appreciate limits were set and communicated and I like the continued trend of easy, family decisions supported by data.

Using data the right way

One thing this pandemic has taught me is that most people have difficulty understanding how to collect, interpret, and apply data to decision making. There simultaneously tends to be an admirable desire to engage with data and a lack of experience to know how to use data. That is most definitely the case illustrated here by our school administrators. The turbulence created from the differences between our family conversations and the decisions being made by our school administrations has been frustrating.

Our school district has set 50 infections per 100,000 people in the population as the limit for in-person learning, and when that limit is crossed, a hybrid or online model is to be adopted. That limit is consistent with the State’s recommendation but differs from the County’s recommendation of 15 infections per 100,000 people, though the limit is not the issue. How to measure whether or not that limit is, or is about to be, exceeded is the issue.

State and County metrics are reported weekly, but with a delay of at least three weeks, so when looking at the information, you have to consider the most current data you are seeing represents what was happening three weeks ago. Because the data is delayed, the direction the data is moving, the trend, must be considered along with the aging metrics in front of you. You may not see the data points for the last three weeks, but the data is still there, waiting to be discovered.

Our school district says in-person learning will be limited when our community sees more than 50 infections for every 100,000 people in our population. Our State is at 67. Our County is at 70. Every area adjacent to ours is above 50, with the nearest two areas at 78 and 80. Our area is reported at 19. Remember, all these metrics are as of three weeks ago. Every one of these local areas has seen rates of infection grow for several weeks in a row. Because no new interventions are being applied, and previous interventions are being removed, historical data suggests the growth of infections will continue and likely accelerate. Considering the trends of infections rising sharply each week in our community, all the surrounding areas exceeding our own school district’s limits, and the available data being at least three weeks old, at what point do you look at the trend and predict our community has already exceeded, or is soon to exceed, the established limits?

COVID response statues from Maricopa County, Arizona where Moderate is recommended to trigger a hybrid learning model and Substantial ends all in-person learning. Multiple local areas are approaching Substantial and trending upward as of three weeks ago. https://www.maricopa.gov/5594/School-Metrics

Much like flowing water, a trend will continue until acted upon. There is no reason to believe the trend of increasing infections will stop or change direction without intervention, so when we see this trend, we should consider taking action to redirect it, even though the limits themselves have not been explicitly exceeded in the data as viewed from several weeks in the past.

All this talk of data is further complicated with news of recent events like the quarantine of more than 200 students from a nearby high school after discovering COVID-infected students as parents and students discussed the open secret that many families had COVID-like symptoms and even verified infections they chose not to disclose, all the while continuing to send their students to in-person classes and school events because they wanted to avoid being stigmatized. Lackadaisical enforcement of on-campus measures like required mask wearing, one-way hallways, and various social distancing policies have also compounded the knowable problem. As a parent, this is all discouraging.

High school students socialize in tight quarters, most without masks, during lunchtime after returning from Fall Break. Even teachers take mandatory mask requirements lightly (notice the teacher in the top left without a mask).

With research pointing to between 50% and 80% of infections being transmitted by infected people who have no symptoms, heading off trends as early as possible becomes even more important as our community and our country sees our third and largest spike in daily COVID cases in eight months.

There is a lot to discover, absorb, and process

The thing that hurts me most as a healthcare data professional is a lack of data. We can not know what we do not measure. Unfortunately, multiple factors including practicality, privacy, and politics have stopped the flow of information from our communities to our decision makers. The lack of timely, complete data combined with inexperience making data-driven decisions has constrained many communities, including ours, placing us at a disadvantage. For now, I will do my best to share our family’s dinner table data science discussions with anyone willing to listen, because all politics are local, and so are all pandemics.

------

Jason Wreath and his family live in Fountain Hills, Arizona. Jason is a 25-year veteran entrepreneur in healthcare and has operated across the spectrum from founding his own businesses to serving as an executive with publicly traded companies. He is an unapologetic lover of data.