The flood of charts, graphs, predictions, webinars, social media posts, and news about this pandemic is overwhelming. Sorting through the waves of information to figure out what’s reliable while dealing with normal 911 calls and making sure you’re not bringing the virus home to your family is a challenge. I reached out to a few data-driven leaders to learn how they are tackling this challenge:
- Alan Butsch, assistant chief, Montgomery County Fire & Rescue Service
- Tim Burns, captain, quality improvement, Montgomery County Fire & Rescue Service
- Scott Dorsey, assistant chief of planning, Snohomish County Fire District 7
EMS1: What’s happening in your system/community regarding COVID-19 at the moment?
Butsch: We are expecting our surge in a week or two.
Dorsey: We had Patient One for the United States. According to some analysis by the University of Washington, we may have hit our peak last week. Testing started so late that any prediction has quite a bit of guesswork in it. Our call volume has been really low for weeks. Hospital volume is at 50% of normal demand.
Do you have a worst-case scenario that you’re planning for?
Butsch: We tried looking at the University of Washington data to anticipate timing of ICU bed and ventilator shortages, but it is difficult to translate it to EMS. So, we’ve been watching New York City and they have been seeing call volume at 150% of their normal call load. We looked at average call volume for us and multiplied that by 1.5 which suggested that we needed 10 more transport units to supplement our base of 42 transport units. Tim did it a little more scientifically, because that’s who he is.
Burns: I did a demand analysis using the FirstWatch tool for a 20-week period that ended with the end of March from 2019. This approximated last year’s flu season. Then I calculated a 150% analysis of the maximum simultaneous demand. I took the results of this and plugged it into Resource Planner and built a schedule. It came up with an extra six or seven units a day. Some of these were 10-hour cars, some eight, and I added four four-hour cars to cover peak times. If needed, we could staff these four-hour units by taking down two engines.
Tell us about your staffing plans
Burns: Right now, we are seeing historically low leave. No one is calling out sick or taking time off. And our overtime signups are really robust. We have plenty of people right now.
Butsch: We added three mobile integrated health (MIH) cars. We started using the pandemic card in dispatch and these MIH crews are handling the pandemic BLS calls. They are assessing the worried well and trying to leave those folks at home. The leave is so low that we are able to staff those cars with straight time.
Dorsey: We started operating MPDS Protocol 36 in surveillance mode a few weeks ago and we have not gone above that level. We did a short test using the Plan, Do, Study, Act (PDSA) cycle on running alternative response options with five pickup trucks responding to those low-level calls. It worked really well and we have them in reserve to be activated if needed.
We are planning staffing by tracking sick leave. We’ve been using the Homeland Security document on anticipated sick leave. They suggest planning for 20, 25, and 40% of your people being out sick, so that’s what we’ve planned for. If we hit thresholds we have plans for which stations or units that we might need to close temporarily.
Tell us about your approach to assembling and tracking data to help you manage this crisis
Dorsey: We track a number of key indicators using statistical process control charts. Looking at this family of measures gives us a view of how our system is doing overall. Looking at them as a group helps us notice patterns and react quickly. I use FirstWatch and a software called QI Macros.
This labor disruption chart tracks the number of employees that call out sick per day.
We track daily call volume and compare it with average call volume using FirstWatch. You’ll see in this chart that we’ve had a recent drop in transport volume.
We also monitor hospital turnaround times. The recent increases are due to the hospitals changing their procedures for receiving patients. They didn’t want us traipsing through the hallways in our PPE.
Lastly, we monitor two COVID-specific indicators, the number of calls that we run that were confirmed by the County to be COVID-positive cases and the number of transports a day where the patient has a temperature over 100.4 F.
I’ve also analyzed the patient care data for these patients. Here’s one Pareto chart that evaluates primary impression for COVID-19-positive patients.
Burns: Some of this system we built just for COVID-19 and some of it was repurposed from things we were already using to track operational and clinical performance. We have always tracked many things monthly including hospital drop time and transport unit cycle time to complete a call start to finish. I just redesigned these to track daily performance. We use FirstWatch and Power BI to build most of our tracking system.
The first thing we did was get FirstWatch to set up a COVID-19 trigger for us based on CAD data and ePCR data. We track this using statistical process control charts and heat maps. When we activated EMD card 36 that asks different questions we started identifying more patients through dispatch. Using control charts helps keep us from tilting at windmills. It’s tempting to overreact to a one-day jump in the data. But because we can see things in context on a control chart it helps us make more appropriate and effective decisions.
We watch hospital turnaround times closely. Recently we’ve noticed much faster turnover times, probably because the hospital volumes are so low. We also track how many beds hospitals have, how many of those they have staff for, and how many are occupied.
We track our normal scheduled leave, our sick leave occurrences, but we also track the number of people who are signed up on our overtime list each day. So far, our sick calls are down and our OT list is full. We staff 307 people a day and we could almost staff half of that from our overtime list.
We watch our overall and our EMS-specific call volume. You can see that the volume started to drop when the stay at home orders were implemented.
We also track specific call types separately to watch for patterns. Our cardiac arrests have held steady but our responses to skilled nursing facilities and other medical calls are down.
Additionally, we track calls for jail facilities, skilled nursing facilities, PPE use and stock. As this pandemic evolves, we will continue to update and modify our systems.
Our colleagues in New York have noticed that overall task time for handling a call start to finish has gone up significantly probably because of the time to don and doff PPE.
As you are looking at the news with international, national and state-level COVID-19 data, keep in mind that nothing will help you lead your organization through this crisis like having a good handle on your own system’s data.
Having a family of measures that track overall responses, COVID-related responses, sick calls for your staff, hospital turnaround times and PPE supplies is a good start. Having a high level, real-time overview of your interwoven vital processes is essential. During stressful times like these, leaders need to be able to sort what’s meaningful out of the massive amount of information. Identifying the data you need, displaying it in a meaningful way, and using what you see to drive your planning and moment by moment decisions will help you get to the other side of this crisis successfully.
Read next: Longitudinally tracking fire/EMS staff exposure to COVID-19
For those of you interested in guidelines and strategies for COVID-19 from the emergency dispatch perspective, Kurt Mills, head of the police, fire and EMS dispatch center in Snohomish County, shared the following.
Snohomish County 911 Center Guidelines Lessons Learned COVID-19 by epraetorian on Scribd
