A client we’ve been working with asked us to help his Cardiothoracic service increase throughput without adding inpatient beds. For this cardiothoracic service, patients go to a specialized unit that currently has a wide variation in census. In extreme situations, elective surgeries are cancelled if one of the beds in this unit cannot be guaranteed on the morning of surgery. Other weeks, this unit has empty beds and underutilized staff with far fewer surgical patients coming into the unit. This variation in census angers surgeons when surgeries are cancelled, costs the hospital overtime dollars when the unit is overcrowded, and frustrates the nursing staff.
We discovered a small but measurable number of these patients account for over half the bed days because they have a length of stay significantly longer than average. Furthermore, these patients are responsible for the majority of variation in census. In fact, if you removed these patients from the population, the census would be relatively level with minimal variability. Stable patient census makes staffing the unit much more predictable and safely allows more patients to flow through.
Upon further analysis by clinical staff, it became clear that these long-stay postsurgical patients can be identified beforehand by factors such as age and co-morbidity. In other words, it’s not a surprise that these patients have a significantly higher LOS and account for a disproportionate number of bed days. Experienced clinicians can look at the medial record and classify patients as likely to be a long-stay patient before the surgery occurs.
Our modelling software is showing staff how this historical pattern of scheduling multiples of these long-stay patients in a single week, followed by a few weeks of no long-stay patients quickly fills this inpatient unit and can block additional patients. We then model scenarios where these patients are evenly scheduled and show how this helps smooth the census in this unit. Like all our simulations and forecasts, we take variation into account and include in the model the understanding that some patients with fewer risk factors might stay more than seven days, and that some patients predicted for long stays in fact only spend two or three nights in this unit.
Because of these predictive simulations, this department is now building a process to identify these long-stay patients and schedule them more evenly. This will help stabilize the census and hopefully enable the service to add patients while reducing the incidence of cancelled surgeries and/or staff overtime.
This might have been the end of the project but the staff kept going. Seeing the disproportionate impact of this small complex population on census, clinical leaders asked why these patients needed to stay so long in the first place. They started looking at the entire care plan for each patient with the goal of determining if they could sequence their CT surgery such that the patient would have a shorter stay. For example, patients going into surgery with stable medication regimes and other therapies & supports in place might not need to stay in this specialized unit as long. The conversation now includes the optimal time to do surgery so the patient has the best chance of an average length of stay. Our software then modeled the effect of reducing the length of stay of this patient population and illustrated the impact on census demand and variation.
This is very exciting to our client. He asked us to accomplish a specific task and now his staff is taking a modern systems approach to healthcare and realizing CT surgery is one stop in an arc of surgical and medical interventions each patient undergoes. This is the type of thinking hospitals leaders need to inspire in their staff. It’s a win to schedule long-stay patients at a regular cadence instead of bunched up. But it’s an even bigger win to convert long-stay patients into more average-stay patients.
The tools and process our platform provides is giving staff and leadership the confidence to embark on a path of improvement. Change didn’t come from a top-down powerpoint extolling departments to work together – change is coming from a bottom-up approach of looking at data in detail and then forecasting how different clinical and scheduling policies will impact their ability to care for surgical patients.