Picture this: You are driving along a 4 lane highway and your passenger asks “Why are you driving so fast?” You glance at your dashboard and see that you are moving at the speed limit and reply “What do you mean? I’m driving the speed limit.” Your passenger says “Oh, it feels faster. Maybe it’s all the cars we are passing. I guess they’re just driving slowly.” Now imagine you are arriving for your evening shift in the emergency department when the dayshift charge nurse is leaving. You ask “How was the morning?” and the charge nurse replies “Terrible. The waiting room was full all day and it never let up. Felt like we were constantly putting out fires today.” Then, when you meet the physician departing you get “Good luck, it’s been socked in all day. I’ve been sitting on my hands all day frustrated staring at a waiting room full of patients with nowhere to see any of them.”
This is a common scenario for many emergency departments. Each of us is driving along in a lane and monitoring traffic in that lane. Teams may be assigned to a specific pod, unit, or cluster of beds, and may be turning over those beds regularly. Other teams may have critically ill patients tying up their beds with little to no turnover, and still others may be caring for inpatient borders and have no capacity to see patients. Each of us, in our own lane, will see a different traffic pattern. Depending on which person you ask, you may have glowing accolades or complete disgruntlement. Additionally, when there is a bottleneck, it may not be apparent where it exists, or why. We have all experienced travel along a busy highway when all of a sudden all lanes of traffic come to halt. The bottleneck ahead is having an effect on your lane, but there is no way to determine what the bottleneck is or how you can improve the situation. Medical decision making is guided by significant data collection and interpretation. Emergency department flow requires the same robust data for decision making.
Historical performance reports are certainly helpful when projecting volume, discussing staffing, and making comparisons after significant changes have been made to patient flow processes. But, these reports are not helpful DURING the shift. When beds are a commodity and volume is high, real-time data is necessary to determine what changes need to be made, now. Imagine driving along the highway without your vehicle dashboard. You need accurate speed information to make changes, you require notifications of system failures to keep everyone in the vehicle safe, and you need the information continuously updated and available at a glance. It is no different working in a busy emergency department. The systems involved are complex and patients travel through multiple phases of care: triage, initial examination, order entry, specimen collection, diagnostic testing, data gathering and interpretation, and disposition. These processes may overlap, but determining which process is a bottleneck requires live data. If emergency departments are expected to react to fluctuations as a driver does to changing road conditions, real-time dashboards are a necessity.
One of the most common questions that arises when building a dashboard is what to include? Although there may be specific items that you need to collect in interval reports, there are only a few key metrics required in a dashboard. Here are some of the most common:
- Patients waiting to be seen
- Hourly arrivals, total and by acuity
- Hourly discharges
- Hourly new admissions (inpatient or obs)
- Currently staffed beds (nursing)
- Current admits waiting for orders (holds)
- Current admits waiting for bed (holds that have orders)
- Current admits with dirty beds assigned (holds)
- Current admits with clean, ready beds assigned (holds)
- Patients waiting for CT study completion
- Patients waiting for CT reading
- Patients waiting for Xray study completion
- Patients waiting for specimen collection
- Patients waiting for lab processing
- Patients waiting for provider disposition
- Patients waiting for nursing discharge
- Patients in the department, by acuity level
This list may seem long, but we have to remember that the purpose of the dashboard is to provide actionable data. In order to make the correct decision, we must measure the correct elements, and have a standardized response for common scenarios. For example:
1. Waiting room census increased the past 2 hours and is now unacceptably high. Simultaneously, shift change occurred, staffed beds went down, patients waiting for beds (holds) increased, and patients waiting for CT increased. What should be the response?
- Perhaps staffing was based on projected volume, which did not account for holds. If on call staff is available, they should be called in.
- A necessary second step is to check in with radiology. Unless a trauma patient is tying up CT for a short time, delays in radiology should have a backup plan for high volume. If it has not been activated, this would be a good time.
- In this setting, increasing provider coverage (even temporarily) is likely the wrong answer. An increase in patients waiting with a decrease in operational capacity due to holds and nursing staffing causes providers to be underutilized.
- Consider moving staff to triage. A physician, nurse, and tech or phlebotomist can see patients in triage and begin evaluations before a patient reaches a bed. In addition, patients can be screened so only those requiring beds are placed in them.
- Activate the housewife surge plan. If there are admissions holding in the ED, find inpatient bed capacity. Senior administration, patient placement, and environmental services directors should orchestrate the identification and turnover of inpatient beds and placement of ED patients.
2. A local sporting event is taking place and waiting room census is increasing. A quick look at the dashboard reveals that most of these patients are low acuity but there is a larger than projected number of arrivals per hour. What should be the response?
- Examination of dashboard data may reveal that all beds are staffed, wait times for radiology are minimal because low acuity patients are not requiring significant testing, and similarly lab is performing well.
- Additional provider staff may be helpful to treat a larger than normal influx of patients.
- Correlating the dashboard with your tracking screen may reveal that all patients in rooms have been seen by providers. In this scenario, additional nursing staff may be necessary to open non-traditional treatment areas like triage rooms or hallway beds for the extra providers to utilize.
There are numerous other scenarios that may arise, but each has a specific response guided by the information displayed in the dashboard. Simply guessing at the solution may yield an improvement, but it would be purely that, guesswork. Empowering your staff to make the correct decision each time requires education, and the dashboard. Given accurate information, a specific response can be formulated, aimed directly at the bottleneck.
Lastly, dashboard information can be utilized outside of the emergency department. Here is a good example of how it can be used system wide (link). This website displays the emergency department performance metrics for metropolitan Adelaide, South Australia. It is publicly available online and demonstrates an integration of departments throughout the system. This allows for triage of patients pre-hospital, to assist in times of critical capacity. It also allows system administrators to monitor performance at multiple locations simultaneously.
Remember, dashboards don’t fix problems, they measure performance and help you identify problems in real-time. Large electronic health records may have pre-built dashboards available. However, there are advantages to third party software. Rapid changes can be made without affecting the EHR production environment. In addition, the dashboard can be displayed in a web browser allowing non-clinical staff to access information without providing access to clinical data. For more on this, consider starting here: CarePulse (No conflict of interest or financial disclosures).