Published in the October 2009 issue of Today’s Hospitalist
It used to be that patients going to the children’s hospital at Nashville’s Vanderbilt University often faced long delays, particularly in the emergency department. Once the Monroe Carell Jr. Children’s Hospital moved into its new freestanding facility in 2004, however, the hospital was determined to get patient flow on track.
Five years later, that process is still ongoing but has produced some very impressive results, according to associate chief of staff Paul Hain, MD. The hospital-wide initiative to improve throughput has drawn not only on a team of physicians, nurses and IT specialists, but also on a number of workflow theories taken from the business world.
“It took about two years,” said Dr. Hain, who also founded the hospital’s pediatric hospitalist program, “to start seeing real results and to get all the information technologies and processes aligned.”
At a pediatric hospitalist conference held this summer in Tampa, Dr. Hain pointed out that stalled throughput not only affects a hospital’s revenue but its safety record, reputation, and patient and employee satisfaction.
He outlined several solutions that the Children’s Hospital used to re-engineer how patients move through the hospital, strategies that can be applied to both pediatric and nonpediatric facilities. Dr. Hain and his team focused on three pieces of the throughput puzzle: accelerating work flow, matching capacity to demand and smoothing out variation.
A broader role for hospitalists
A key part of improving throughput at the hospital was to simply use its hospitalists more, particularly when the census was high and bed demand far exceeded supply. When bed space is at a premium, Dr. Hain said, hospitalists not only round on their own patients, but ask residents and charge nurses what other patients may be ready for discharge but are missing discharge orders.
Hospitalists then call the attendings or surgeons of those patients, he explained, “and they’ll actually push people to send kids home who are ready but aren’t leaving through sheer inertia of the system.” Hospitalists, who are employed by the medical school, offer to go see those patients and not bill for the discharge. “That’s just part of our job,” Dr. Hain explained.
“In our hospital,” he added, “we call ourselves the floor sweepers. We help decrease length of stay by pinch hitting in trying circumstances.”
That approach has also improved throughput in the pediatric ICU, Dr. Hain said, because hospitalists never refuse a transfer request from the PICU, particularly the “really sick kids” that would make some general pediatricians nervous. That spirit of cooperation works both ways.
“The converse is that if patients get really sick again, we just send them back, and the PICU doesn’t give us any trouble,” Dr. Hain said. “The PICU guys are our biggest supporters in the hospital.”
Better bed management
Expanding the role of hospitalists, however, was only part of the solution. How do physicians know, for example, which patients they need to step in for and discharge? To answer that question, the hospital created a new position: the administrative coordinator.
“It turns out that in a hospital this big,” Dr. Hain pointed out, “it’s really hard for people to know where all the patients are and to solve all the flow issues unless that’s their only job.” Administrative coordinators are nurses who have worked at least five to 10 years in multiple units of the hospital, he said. Each now works a 12-hour shift, with coordinators covering the hospital 24/7.
Administrative coordinators are the go-to people for any question or problem, from “I need a car seat for a kid to go home” to “Can this kid with an infection share a room in the ED?” Even more importantly, coordinators are charged with watching the “bed board” “which tracks all of the hospital’s 220-plus beds “”to make sure things are moving the way we think they ought to be moving,” Dr. Hain said.
Coordinators send out updates on bed conditions every 12 hours that include information on how many patients are in each hospital pod, how many are in surgery, how many discharges are anticipated and any other information that might affect bed availability.
“This lets everybody involved in bed management keep up on a 12-hour basis,” Dr. Hain said. “It’s amazing how many fewer phone calls there are because everyone already has the answers to most of their questions.”
The bed board itself is another innovation compliments of Vanderbilt’s IT department, Dr. Hain said. This software program, which is available on every computer in the hospital, gives a colorful snapshot of the entire hospital in real time, showing what beds are full, available, in flux or closed. All bed requests from nurses throughout the hospital are made electronically to this central site, and staff can look at the bed board’s electronic queue to see where their requests stand.
The bed board also allows dispatchers to page the cleaning crew when a patient is being discharged. According to Dr. Hain, “The cleaning crews are actually waiting by the room as patients are packing to leave.”
Centralizing transfer calls
Bed management also improved with another throughput innovation: processing all transfer calls in one access center. When another hospital, ED or private physician calls to send a patient, they are directed to that center. (All emergent admissions have to go through the ED, Dr. Hain said. For safety reasons, unscheduled emergent admissions are no longer allowed to go directly to the floor.)
The access center collects all patient information, enters it on the electronic bed board and forwards the history to the physician who will see the patient in the ED. Dr. Hain pointed out that the centralized system has cut down dramatically on one of the top complaints the ED used to hear from referring physicians: “I can never talk to anybody in the ED when I want to call my patient in.”
In addition, he said, requiring all emergent patients to go through the ED rather than being admitted directly to the floor has helped reduce length of stay, in part because children’s conditions can change so quickly. A child with asthma sent to the hospital by a general pediatrician, for example, may not need to be admitted at all after being seen in the ED, receiving three treatments and being put on steroids to take at home.
“We can actually intercept a lot of kids and really reduce length of stay,” Dr. Hain pointed out. “If we send them to the floor, that’s at least 24 hours before those doctors would come to see them.”
Untangling ED bottlenecks
Like many other hospitals, the Children’s Hospital at Vanderbilt discovered that some of its biggest bottlenecks occurred in the ED. Once team members analyzed how patients were processed in the ED (see “Knowing when to ‘elevate your constraint’ “), they realized they had to come up with a faster track for treat-and-release patients.
Before the throughput initiative, patients who were simply treated and released “not admitted “were spending significant amounts of time in the ED. In the winter, the average wait time was 170 minutes; in the summer, that number dipped to 110 minutes.
Team members found that a “fast-track doctor” was in the ED only from 4 p.m. to midnight during the week and from noon until midnight on weekends. At all other times, patient waiting times skyrocketed. (The team also found that the busiest time in the ED with less-than-emergent patients was dinnertime.)
“We decided that we were going to see all treat-and-releases in 60 minutes from the time they arrive to discharge,” Dr. Hain said. Just as importantly, he added, the goal was to reduce waiting times with no change in patient safety, readmission and satisfaction rates.
The hospital has reached those goals in part by adding a fast-track nurse practitioner from 10 a.m. to 10 p.m. It also brought in an experienced ED nurse who triages patients (“with really impressive accuracy,” Dr. Hain noted) to quickly identify who is sick and who isn’t. Patients at that point are sent to either the fast-track or core queue in the ED.
Although these staffing changes initially cost money, they have become cost-neutral over time. That’s because better fast-track care allows the ED to work with fewer nurses, who are no longer distracted by treat-and-release patients.
Throughput was also improved by getting the IT department to write a computerized discharge order set and a “discharge wizard” tool that guides quick production of a discharge order, home action plan and prescriptions. “It’s almost like checking out of Amazon,” Dr. Hain said.
Another fix was to create a discharge lounge where patients who have been discharged but are waiting for a ride are moved so their bed can be given to someone else. Although the lounge can’t be used for contagious patients, it does free up some much-needed bed space.
The next major bottleneck to untangle is smoothing surgical throughput, especially in scheduling elective surgeries. Those surgeries tend to cluster on specific days.
“Medical elective admissions are also grouped on particular days,” Dr. Hain pointed out. “Convincing physicians to spread that work out over the full week would dramatically help throughput.”
Deborah Gesensway is a freelance writer who covers U.S. health care from Toronto.
THE SECRET TO BETTER hospital throughput is to identify “and resolve “bottlenecks in the system, according to Paul Hain, MD, a pediatric hospitalist and associate chief of staff at the Monroe Carell Jr. Children’s Hospital at Vanderbilt University in Nashville. To do that, Dr. Hain learned a lot from workflow theories that come from the business world, complete with some business-speak.
The first of those is known as “the theory of constraints,” Dr. Hain said during a presentation on throughput at a pediatric hospitalist conference held this summer in Tampa. The initial step is identifying “the constraint,” which means a bottleneck or back-up.
First, analyze the entire process of care, breaking it down into all its steps and personnel. The biggest constraint might be something as simple as the fact that the ID printing machine takes 40 minutes to print a label, Dr. Hain explained, “so everybody is sitting around waiting for these labels to come out and nobody can get any work done.”
After identifying the constraint, it’s important to “subordinate all other processes.” In this case, that could mean ignoring complaints about other problems until you figure out how to make sure the ID printer is never left idle.
The third step is to “elevate the constraint,” meaning that you allocate capital to remove the bottleneck. In this example, said Dr. Hain, that could mean buying another ID printer.
Finally, if you still have back-ups, “go back and start over because your constraint is something else,” Dr. Hain explained. “You repeat this cycle until you’re going as fast as you really want to go.”
One common mistake is to jump on the first bottleneck you identify when there may be an even bigger problem further down the line.
You may find, for instance, that ED triage is pretty efficient, but registration can process only three patients an hour. That back-up needs attention, but an even bigger problem is that physicians, further down the chain, can see only two patients an hour. Even if you fix the constraint at registration, “your output will never change,” Dr. Hain pointed out.
Another approach for improving throughput is queuing theory, a methodology that matches variable demand to fixed capacity. Queuing theory can help explain, for example, “why variability isn’t always a variable.” Consider ED demand, Dr. Hain said, which is more predictable than people realize.
“You never know when the trauma is coming in,” he noted, “but you do know what times of day are going to be busier.” Queuing theory suggests that you should staff up during those times and down during others. “That might sound self-evident,” he said, “but it doesn’t happen in many places.”
Dr. Hain said physicians should consider another approach pulled from manufacturing known as “lean processes.” Like Six Sigma, lean processes are a method of identifying “waste” ” unacceptable variation “based on a system of sampling to identify a step in a process that may be broken.
“It’s a way of using the variability in a small subset to describe the large set, and it works very well for processes in a hospital,” Dr. Hain said. “You may say, ‘No, people are different from ball bearings,’ but it turns out they’re not quite as different as you might think.”