We dare anybody to find an area of healthcare faster paced, stressed, difficult and anxiety-provoking than an emergency department (ED) these days. Within the ED there are constant and never-ending inputs and outputs with more process ingredients being added to the “cooking pot.” To prevent this ED pressure cooker from blowing its lid, emergency department managers require real-time and retrospective analytics with the goal of relieving the four key types of pressure. In the ED, these are:
•Clinical quality pressure
•Service and timeliness pressure
•Staffing and skill deployment pressure
The authors of this article have worked independently for the last two decades on healthcare delivery, analytics and business intelligence development projects. We have helped many organizations learn how to apply analytics to make their work lives better, help them do their jobs more efficiently and become more profitable. Despite these independent paths, we have found many commonalities and want to share them here.
For purposes of this article, we have broken the above four pressures down into the key business questions that need to be dealt with, and show how analytics has been used to effectively reduce departmental pressure. At the end, we will show the reader how to tie these analytics all together, and further demonstrate that using business intelligence is essential to not only relieve today’s reporting requirements, but to get your ED ahead of the curve.
Four Key Pressures in Managing the ED
Healthcare providers, including acute care, ambulatory care and ancillary care are embracing a model with a triple aim, which includes quality, service and cost. Progressive organizations have added a fourth “aim,” making the mission a “quadruple aim.” This fourth aim centers on staff, physician and partner engagement. Since providing healthcare is a service business, this fourth aim in many ways ties all of the others together. The ED ties them together at a rapid pace, and is therefore a propitious place to begin your business intelligence journey, or to improve it once you have begun. If your analytical capabilities can make it there, they can make it anywhere.
However, before you begin your journey of collecting, analyzing and presenting actionable information, there are some key business intelligence principles you should consider:
Data is everywhere and can get overwhelming. Establish a list of 10-12 key performance indicators (KPIs) and focus on metrics that can be directly impacted by a behavior change.
When reporting KPIs, application of the balanced scorecard approach and its principles of showing results as red (your threshold value), green (your target value) and yellow (caution level) is a best practice. If you require a starting point for target and threshold values, calculate and use 2 times standard deviation above or below the mean for each KPI.
Don’t just report KPIs, but rather create drill paths from KPIs to actionable data. This drill information may include lists of patients outside a threshold by physician/mid-level practitioner (MLP), by nurse, by registration clerk, etc. Use these lists as a feedback mechanism for KPI improvement.
Start by focusing efforts on KPIs that are red. But do not stop there. Those in your organization with consistent green metrics are your teachers down the line.
Application of the above best practices, how these KPIs apply to each of the four above pressures and how those measures are used to manage that pressure are detailed below. Your organization will undoubtedly have other measures of interest and other uses for those measures. As a matter of fact, we hope that this list serves as a mental springboard for you to say, “Oh yes, we could also use that measure to improve X and Y and Z.”
If you want more doctors and nurses in your institution to care for your increasing workload, you will obviously need to find a way to pay for these resources. A first place to look is at some “low-hanging fruit.” The authors’ experience in reviewing financial performance within EDs has proven a great deal of revenue is often left on the table because of a disconnect between systems that document care and service delivery, and systems that document what is actually billed (the specific CPT codes). By comparing this information at a patient level or in the aggregate between what is provided and what is billed by physician, nurse or medical coder/biller, outliers can be identified and education can be initiated. This could involve educating a specific physician or a group of physicians. It could involve educating one biller about coding or educating the entire group of billers. It could also involve enhancing how the electronic health record (EHR) allows physicians or nurses to document more effectively so billers can code more appropriately.
Here are some key financial measures to analyze on an aggregate or physician-specific level:
Total Monthly Charges. This is the ED’s primary financial activity measure, which ties together demand patterns and revenue patterns.
Average Payment per ED Visit. When you got paid, what was your average payment, including write-offs and no-pay? This metric is most important for cash flow projections and budgeting. However, due to delays in payment, it may take 3-6 months from date of service before you get a relatively complete result for all patients seen and billed in a particular month.
Average Work RVU per ED Visit. Relative value units (RVUs) are tied to CPT codes billed and indicate complexity and amount of patient care services. This metric has become very popular in many physician incentive programs. If a physician frequently bills a higher E&M (evaluation and management) level or provides more than average ancillary billings, that physician’s RVU per ED visit will be higher than his/her peers. Production of this metric between physicians may flush out those physicians who are not billing appropriately.
Appropriateness of E&M Coding. E&M coding is often a telling statistic. Most billing entities provide E&M code percentage reports as routine reporting. However, a more meaningful metric would include production of the percentage of patients admitted to the ICU (intensive care unit), CCU (critical care unit) or telemetry, billed at a Level 5 or critical care code. Evaluation and improvement to this metric may be some of the low-hanging fruit identified for revenue enhancement activities. Identification of cases not billed as a Level 5 or critical care and review of note documentation may create the learning your group needs to improve its E&M coding performance.
Clinical Quality Pressure
Most of what you do today, in the area of analytics, is probably focused on the production of quality and other metrics required by Medicare and Medicaid. You may already have teams of people focused on the production of “Core Measures,” “Clinical Quality Measures,” or “PQRS Measures.”
Here are just some (not a complete list) of the key clinical measures you are probably already reporting:
•Aspirin at Arrival for Acute Myocardial Infarction.
•Stroke and Stroke Rehabilitation: Deep Vein Thrombosis Prophylaxis (DVT) for Ischemic Stroke or Intracranial Hemorrhage
•Community-Acquired Pneumonia (CAP): Empiric Antibiotic
•Median Time from ED Arrival to ED Departure for Admitted ED Patients – Overall Rate
But what the authors have found is many institutions produce these metrics but do not provide drilldown information by physician and mid-level practitioners because it takes too much time to produce and distribute reports at this specific level. Therefore, it is not surprising that metric improvements with “Core” and other measures are not immediate. Similar to financial metrics, production and presentation of clinical measures by physician or MLP requires feedback if improvement is desired. Practitioners must be given information to act and must have accountability for acting. There has been a lot of work over last 10-15 years on defining these measures, but little has been done on defining how a practitioner can act on those metrics.
Service and Timeliness Pressure
If you ask patients what they dislike most about visiting an emergency room, it would be the wait: the wait for a room, the wait for a doctor and the wait to get home or be admitted when all clinical activities have ended. Therefore, it is as just as import to focus on these time increments as it is to understand and report total turnaround times.
As with financial and clinical measures, this information must be produced at a physician, nurse and MLP level. These metrics must also be further segmented by emergency department area – fast track versus acute, and by time of day and day of week.
Here are some key service and timeliness measures to analyze:
Patient Walkout Statistics. This measure is presented as a whole number and as a percentage of total patients, and includes patients who left prior to or after triage, but before they were seen by a physician. Having a patient leave your ED is not only a potential financial lost opportunity as well as a clinical lost opportunity, but has a huge impact on peoples’ perception of the quality of your healthcare services. There is also a practice risk associated with patients who leave before they are seen. A best practice is to drill down on this metric by time of day and day of the week. Also, look to see if trends exist and if there is a statistical difference when specific nurses are working in triage. Also compare this statistic to patient length of stay or the sub-metrics that make up the overall length of stay to see if causality exists.
Door to Room. Patients have a tendency to mentally relax once they have been placed in a room. This has a clinical impact in calming down patients’ anxieties, making interventions easier to perform. It has a service impact too because once in a room, patients are less likely to leave. It also has a financial impact because revenue does not walk out the door. A surrogate measure is door-to-practitioner, which has a similar positive impact. Monitor this measure closely and act on what you find, good or bad. Also drill down on this metric by time of day, day of week, triage acuity, the patients’ final diagnosis and ED area.
ED LOS for Patients Discharged. Length of stay (LOS), also known as turnaround time (TAT), for patients who are discharged measures the length of time between when a patient arrives and when a patient leaves the ED; in total. This metric is either reported as a mean or as a median value. To determine the root cause of poor performance with this metric, it is essential to drill down by doctor, by time of day, by day of the week and by ED area. The reader may also want to exclude from this calculation patients who artificially elongate this measure, such as psychiatric patients or patients placed in an observation status. Finally, the reader will want to analyze and monitor how much lab or x-ray turnaround times have had an impact on the overall length of stay.
ED LOS for Patients Admitted. This corresponding measure, turnaround time for patients who are subsequently admitted to the hospital, is equally as important. A subset measure is physician decision to patient admitted. By producing both, the reader of the reporting will determine how quickly decisions were made and by what practitioners. This metric is also either reported as a mean or as a median value. To determine the root cause of poor performance with this metric, it is essential to drill down by doctor, by time of day, by day of the week and by ED area, by inpatient location, etc.
Staffing Deployment and Skill Development Pressure
As mentioned in the introduction, staffing deployment/skill development is the category of measures that represents the intersection of clinical, financial and operational metrics. Healthcare is a service, and people provide that service. Because it is a pivot point for the other types of measures, staff deployment and skill development must be supported through feedback mechanisms, which often consist of comparative statistics between practitioners or nurses. Subject areas can include financial/billing, clinical quality, service and satisfaction. These are collectively referred to as productivity metrics.
In evaluating productivity metrics that are given the most weight, the authors have surmised that too much emphasis is placed on a practitioners’ financial performance (RVUS per hour) and not enough on operational, clinical and satisfaction performance. In discovery, it is determined that those who wish to produce productivity metrics are often limited by what data they can collect and match at a practitioner level, especially satisfaction.
As a best practice in creating productivity analytics, the authors have found presentation of physician names, rather than the use of codes, is more impactful. Further, it is essential to only produce those metrics on a productivity report that can be directly impacted by the person or group being analyzed. For example, physicians may not be able to impact how long it takes to place a patient in a room, but they can impact how long it takes to see a patient once they are in an ED room. Also compare individual performance to group means and medians and highlight examples where individual performance is below the group performance. Finally, remember your objective is to improve, not to penalize people for poor performance. So also highlight good performance as well as improvements in performance.
Here are some examples of key metrics that may be part of your productivity analytics:
Average Raw Patient Satisfaction Score by Question Group. The type of feedback that each practice is getting and comparisons to what the ED as a whole is getting are important to understand what doctor sets and/or nurse sets are contributing (or hurting) patient satisfaction. Depending on whether you are measuring nurse or physician/MLP performance, it is important to be able to segment your survey results to reflect only those questions asked about the group you are measuring. As mentioned above, satisfaction has huge financial, operational and even clinical impact.
ED Room to Physician Decision. For patients who were admitted or discharged, how long did it take the physician to make the decision?
Patients Seen by Physician and Provider Hour. This metric is a result of total patients seen as a ratio of total hours worked.
RVUs per work hour. This expands upon RVUs per patient and takes into consideration the number patients seen per hour.
Telemetry Procedures Orders that are Subsequently Billed. This metric gets to the heart of unbilled procedures and revenue recovery.
CTs Ordered per 100 Discharged Patients Seen. Another clinical metric that looks at the use of CTs as a diagnostic tool.
Controlled Substances Ordered per 100 Discharged Patients Seen. Another clinical metric that looks at the use and or abuse of controlled substance ordering habits.
Getting Ahead of the Curve
As you can see from the last list of metrics, these measures cross the four borders. Getting ahead of the curve means deliberately defining measures and delivering information that links as many of the four types of needs as possible. This makes managing them more challenging because relieving one type of pressure can cause an increase in another type of pressure. For instance, one hospital was looking at critical care financial measures. They were substantially under-billing patients. With further analysis, the hospital found great variances across physicians, which they presented at staff meetings. This was caused by specific coders who were not interpreting charts correctly, either as a result of physicians not communicating the correct codes effectively or from a gap in coder knowledge. Education of coders and physicians resulted in additional total collections in the neighborhood of a quarter million dollars. This example crosses financial, staff engagement and even operational measure categories by giving both groups feedback on how their actions had dollar, time and quality impact.
Another instance was a hospital with a walkout percentage hovering at 2.5%. They were able to reduce it by reducing turnaround time per patient by 30 minutes. Additionally, this reduction in turnaround time impacted overall satisfaction scores, increasing the hospital ED’s percentile performance from mid 70% to over 90%.
The strategic objectives of your organization help guide which pressure to work on most, second and so forth. Is financial success (or survival) your number one priority? Clinical quality performance? Operational survival?
In addition, the maturity of your organization’s analytics capabilities further helps to guide your business intelligence development efforts. For instance, does your organization have a strong clinical quality measurement team? This is a skillset to tap to make your ED analytics capabilities shine. Does your organization make use of process engineering, lean methodology or service gurus? These could be fruitful areas to apply to the ED.
Choose an area where you have the most pain to alleviate or the most gain to be garnered, and begin there. ED analysis is the place where you will find out fastest whether to go ahead or to change gears.
Get started now. Talk with your peers in other organizations to see what works and what doesn’t in the area of ED measurement. Talk with your vendors and consultants. And, of course, talk with your management to understand the strategy regarding the ED, as well as people right there on the floor in the ED itself.
We hope that our shared experiences help you to get started, and hope that your efforts in developing ED analytics make a difference.
Thanks for reading!