House and Senate leaders rushed to pass a bill last week that allocates $16.3 billion to overhaul the Department of Veterans Affairs’ troubled healthcare system.
Delayed care and falsified documents have put the spotlight on a critical physician shortage and an outdated scheduling system that cannot accommodate a growing patient population. The bill will provide funding for some patients to receive care from government-paid private doctors while also boosting the VA’s own physician headcount, if modestly.
What it won’t do – at least not immediately, say anonymous sources – is upgrade the department’s scheduling system, which the acting Veteran Affairs secretary, Sloan Gibson, says is critical. Instead, it will create a task force to look into the problem, but it does not appear to provide the funding to fix it.
“We need to focus on increasing efficiency at least as much as adding resources,” insists Kumar Muthuraman, associate professor at the McCombs School of Business, who researches and develops healthcare scheduling models. “It’s a no-brainer: If you add new staff without ensuring that existing resources are optimally utilized, you could have doctors staring at the ceilings some days while patients continue to wait weeks or months for appointments.”
Head Count Doesn't Equate to Efficiency
And because there’s a nationwide physician shortage, the VA, which pays its doctors less than they would make in private practice, could have a hard time filling its 400 vacancies for primary care doctors. Having clinics that are adequately staffed is paramount, but making sure each doctor’s time is properly managed is crucial for administering efficient care.
And that’s where Muthuraman and Professor Doug Morrice, both of the department of Information, Risk, and Operations Management, come in. They’re developing models to help healthcare systems, like the VA, improve scheduling. By analyzing data that clinics and hospitals already collect on patients, they can slash waiting times while simultaneously enabling doctors to see more patients each day.
What might be most intriguing about their work is that neither professor has a background in healthcare. They’ve worked in fields as different as finance and oil exploration.
Morrice is a systems engineer who’s consulted with companies from Schlumberger to PricewaterhouseCoopers and Texas Instruments.
Whatever the line of work, Morrice looks at it as a process. “Processes are everywhere,” he says. “It doesn’t matter if you’re in an oilfield, a retailer in a warehouse, or a factory with computers. A process has many steps, and it has bottlenecks.”
His aim is to coordinate those steps in the most efficient way possible. “You’re scheduling a particular piece of equipment to show up at a seismic job at a particular time,” he says. “Or you’re scheduling a doctor to show up at the operating room at a particular time.”
At the request of the University of Texas Health Science Center at San Antonio, Morrice looked at the growing field of outpatient surgery. He worked jointly with professor Jonathan Bard of the Cockrell School of Engineering.
They found that the key link in the surgical chain was the anesthesiology clinic. Its goal was to assess 100 percent of patients a day or two before surgery, making sure potential problems were detected before patients got to the operating room. The trouble was that the clinic maxed out at 40 percent.
Many patients, the researchers determined, could be screened over the phone. For those who needed to be seen in person, the number one cause of delay was missing information. Explains Morrice, “The clinic would have to make multiple phone calls and searches of electronic medical records to get the information together.”
His solution was to put a single person – a registered nurse – in charge of compiling that information. With help from two new medical technicians, for an extra $50,000 a year, he estimated the hospital could save $1 million, from reducing surgical delays and scratching unnecessary tests.
The new system went live at UTHSCSA in August, 2013. Since December, reports Morrice, patient loads have been steadily rising. In May, the clinic assessed a record 423 patients, up 22 percent from a year before.
Decrease Wait Time and Reduce Costs
Elsewhere in McCombs, Muthuraman approaches healthcare scheduling from a different background: finance.
He says the dynamics of how someone decides to trade shares of stock are surprisingly similar to how a clinic schedules patients. Both problems revolve around making decisions under uncertainty: about stock prices or whether a patient will show up for an appointment.
“The biggest problem in clinics is no-shows,” says Muthuraman. “You schedule two people, and sometimes both don’t show up. The doctor is sitting for 20 minutes doing nothing. So you schedule three people, with the anticipation of one canceling. In the morning, all three show up, and you’re overbooked.”
Instead of offering every caller the first available appointment, he says, a hospital should try to balance its risk of no-shows. By analyzing patient histories, it can predict how likely each patient is to actually arrive. The less likely they are to show, the more of them can be booked for the same slot, with a small risk that they’ll all show at once.
With the help of doctoral student Ester Wang, Muthuraman has devised a computer model to match patients with time slots. When a patient calls, the computer suggests to a scheduler the three optimal slots for that particular patient. Says Muthuraman, “You can maximize the number of patients you see and minimize the waiting time each patient goes through.”
He wrote the model for a UT supercomputer, but he’s working on a simplified version that can be run on a desktop at any doctor’s office or hospital. Preliminary tests find it to be only 2 percent less efficient than the original. Now he’s trying it out at UTHSCSA and seeking a federal grant to commercialize it for systems like the VA.
Despite the VA’s well-publicized problems, it has one thing going for it, says Morrice. It already collects the kinds of big data these computer models need in order to churn out useful results. The missing link, he says, has been the ability to use that data to improve operations.
“One thing I often find with electronic medical records is that they they’re well thought out for billing purposes, but they aren’t as well thought out on the operations side of things,” he says. “Good data is the key factor for better operational performance. If you don’t have good information, you can’t do this.”
And good data that’s already been collected is an investment no one’s using. “Upgrading the scheduling system would be way less expensive and easier compared to finding physicians and nurses to hire,” says Muthuraman.
- Additional reporting by Adrienne Dawson
*Article has been updated to reflect final dollar amount of approved legislation.
The McCombs School of Business is dedicated to educating the leaders of tomorrow while creating knowledge today that shapes industry, society, and policy. Ranked 5th in the world for research, McCombs has one of the most productive faculties of any school, public or private. We educate more than 12,000 students each year and instill in them the desire for intellectual inquiry and integrity that defines The University of Texas at Austin.
At McCombs, discovery shapes learning and practice.
Photo by Terry Ross / Flickr Creative Commons