Maximize MR Throughput with Efficient Scheduling

As little as one extra MRI per day can generate more than an additional $200,000 in incremental revenue annually. But most imaging centers use crude scheduling systems that do not accurately present a center’s potential throughput. David A. Dierolf, director of performance improvement, Outpatient Imaging Affiliates (OIA), Nashville, Tenn, outlined a handy method for understanding the potential of your schedule to enable maximum throughput for an audience gathered at the May meeting of the Radiology Business Management Association in St Louis, Mo. Hired by OIA in advance of the Deficit Reduction Act to improve efficiency throughout the company’s imaging center holdings, Dierolf, an IT expert, shared two case studies with the audience that revealed his techniques and yielded significant additional incremental revenue at the centers. OIA specializes in establishing joint ventures with local health care providers and operating those centers for its partners. Attributing OIA executive VP, operations, Kelly Gill, as the inspiration for his talk, Dierolf said: “One of the first things he told me was nothing is worse than unsubstantiated success. I am going to build my talk around that.” University of Virginia Imaging This very busy freestanding imaging center in Charlottesville, Va, offers full diagnostic services, adding up to 6,500 exams per month. A partnership between OIA and the University of Virginia, the center operates four very busy MRs and is already working towards a fifth. “Enhancing utilization is important when your schedule is full,” noted Dierolf. “If you are only using half of your slots, there’s not a lot of reasons to spend a lot of time trying to get one more procedure through unless you are going to send you techs home halfway through the day.” The center’s third available appointment was 8 or 9 days out in the fall of 2005 when Dierolf entered the picture. Dierolf was charged with developing a method to assess whether the scanners were running at maximum capability and also to develop a better method to decide when a new scanner should be purchased. Dierolf had to determine the following measures: Throughput, or how many exams were done in a specified time period, one week in this case. The number of appointment slots, defined as a unit of time reserved for a patient or an exam. Because the existing scheduling system dated back to 1970, there was not a lot of flexibility. “Back then we were scheduling everyone into 45 minute time slots,” noted Dierolf. “There was no ability to do some at 30 and some at 45. When scheduling and deciding what is the next available appointment, you looked at 4 different screens. There were three high field magnets and we kept those open in 19 slots a day, M-F, and 12 on Saturday. We also had one open system. The protocols were longer and the scanners were slower, so we scheduled those at 60 minutes.” Dierolf counted 371 total appointment slots per week, but still did not know how many exams could be done in a week. “That question was difficult to answer,” said Dierolf. “We knew how many patients we saw and how many exams we did, and we even tracked the number of no-shows and last minute fill-ins. But we couldn’t figure out how many slots we did we not use. It was the number we needed to know in order to find out how many exams we could do.” Calculating Theoretical Maximum Throughput To answer that question, Dierolf had to do two things: figure out how many exams were done in a specified number of appointment slots so he could calculate how many slots the center wasn’t using: “It was interesting that we really couldn’t figure that out.” Dierolf undertook the painful method of going to the logs of each machine and poring over the notes that the technologist kept on the schedule: did not show, or “x” for cancel. He transferred those notes to a spreadsheet. “Today we actually have an online system where they can go on and say here is the schedule and here is what happened and it makes it fairly easy to count, “ he noted. He performed the task for the month of October and went back and did it again in December in his effort to determine how many exams he could get into each appointment slot. Because MR of the brain and MRA of the head could be performed in one 45-minute slot, knowing the patient number was not enough because sometimes two procedures took one slot and sometimes two procedures took two slots. “This notion of coming back and evaluating the procedure mix proved to be key, and this is probably was one of the things that I discovered that made it much easier to get this information for other modalities or other centers,” Dierolf explained. Dierolf recommends following this procedure to avoid having to go through the appointment logs: First identify exam sets of two exams done on the same patient during the same visit. Then determine, for these exam sets, how many appointment slots are scheduled And, finally, calculate how often those exam sets occur in a month Table 1. Expected Exams per Slot For instance, exam set 1 (see table above) was MRI of the abdomen, MRA of the abdomen. So whenever that exam set occurs, Dierolf saved one appointment slot. Because that exam set occurred 78 times in a month, he knew he could save 78 time slots in a month. He repeated that exercise for every exam set the center performed. “Sometimes like for MRI brain, MRA head, and MRA neck, three exams fit two slots,” he explained. “Sometimes I don’t save anything: MRI brain, MRI C-spine, two and two, and I don’t save any slots.” Dierolf went through this process for all of the exams for one month of data, and found that he performed 1295 exams, with a savings of 127 slots, that used 1168 slots. To find out how many exams he gets for each time slot, he divided 1295 by 1168, to discover that he gets an average of 1.11 exams for every appointment slot. “So a key component is the only way you are going to gain this is if your schedulers know anytime I’m doing a MRI of the brain and an MRA of the head, schedule that in one chunk, and that’s an important efficiency gain,” Dierolf explained. “By doing the analysis, you know what common exam combinations that you have, and this is a really important number, it’s a driver: 1.11 exams per slot. “This is a good 10% 11% savings, so if I am scheduling one exam per slot, I am off 10% right from the beginning.” When you know your expected exams per slot, you can then determine your maximum throughput figure. Dierolf cautioned that this calculation is theoretical not practical and, center managers need to understand this is a tool to gauge efficiency, not necessarily a realistic goal. So, in the case of UVI, which had 371 slots per scanner per week, with the potential to do 1.11 exams per time slot, managers should not necessarily expect staff to push through 411 exams in a week. “So, we know what we can do, and the next question is, how are we doing?” Dierolf explained. “It is really a measure of how well a center is using the schedule. I take what I am doing divided by what I could be doing: 73% is where we were.” Figure 1. Algorithm for calculating theoretical maximum throughput. Taking Control Once he had a measure of efficiency, Dierolf turned his attention to a list of variables he called efficiency detractors that could impact exam throughput: No shows Last minute canceled or rescheduled Unable to complete Isolation (VRE/MRSA cleaning) Equipment downtime/planned maintenance Unexpected staffing problems Unscheduled appointment times In investigating the process for appointment confirmation calls, he discovered that the calls, made 48 hours in advance, were generating the problem. “Those early measurements showed that even with that 8- or 9-day third available appointment, we had three unscheduled slots per day,” Dierolf noted. “We were calling people 48 hours in advance and they were saying that’s right, I can’t keep that appointment, and canceling and rescheduling for a day or two later. We weren’t being efficient at refilling, reusing that slot, so we were generating a lot of churn. The schedule was full if you looked out, but when we looked at tomorrow, there were always available appointments.” The solution was to dedicate one of our schedulers to “make tomorrow full.” That scheduler opportunistically looked at the schedule for patients that were 3 or 4 days out who would not cause pre-authorization or pre-certification problems, and try to switch them to a next-day appointment. That gave the regular schedulers time to fill those later appointment slots. “We were actually able to move up 1 and a half patients a day,” said Dierolf. “So we cut that churn we were causing almost in half.” UVI also: Added evening and weekend hours, including Sunday, for which there were some associated costs. Replaced an open magnet with a 60-minute time slot with an open magnet with a 45-minute time slot. Squeezed appointment times on two of its four machines from 45 to 40 minutes, adding two additional appointment slots per machine per day, or 20 a week. To ease the concern of the technologists, UVI blocked one of those extra slots per day, so every day there would be an unused slot that could be used to catch up if the technologists fell behind for some reason. Ultimately, UVI increased appointment slots to 407 a week, the efficiency has boosted to about 82% and they are averaging around 433 exams per week. At the same time, the third available appointment was reduced to five days. Raleigh Radiology Raleigh Radiology, Cedarhurst, NC, is a privately held freestanding imaging center that offers a full range of services, with an examination volume of about 2,600 a month. Because North Carolina is a certificate-of-need state, Raleigh Radiology works with a contracted MR, 6 days a week for 12 hours per day, but volume demanded that they run 15 hours on weekdays and 8 hours Saturday and Sunday, resulting in considerable overtime costs on the contract. Appointment slots were fixed at 45 minutes and the administrator knew there was slack, but he liked the flexibility of being able to take add-ons. The question Dierolf needed to answer was this: how many examinations could be done in 30 minutes? Could some idle time be removed and still allow time for add-ons? Dierolf worked with the chief technologist to identify which exams could be done in 30 minutes: primarily no-contrast extremities and some spines without contrast. When added up, it was determined that 74% of their exams could be done in 30 minutes. Table 2. Raleigh Radiology Exam Mix Raleigh began slowly by choosing several to be scheduled at 30 minutes—the cervical spines and shoulders and thoracic spines—thereby easing into the new schedule with 30-minute appointment slots. They were all scheduled in the evening, because they did not have contrast coverage in the evenings. “One of the challenges you have when you are working with different time slots is you can end up with holes, 15 minute slots, and I don’t have any 15 minute exams to stuff in there,” Dierolf noted. “So the scheduling folks worked hard to pack the people in and in some cases they had to call people up to say we scheduled you 15 minutes early. It took some work. What we ended up with is getting more exams per day, and since we were getting more exams per day, we could reduce the overtime we were paying on that contract. So we actually reduced the cost of the exams.” Dierolf shared a tool he used to plot exam times and thereby measure success. Called a Control Chart, the y axis showed the number of exams per hour, and the x axis represented time. Each exam is plotted, providing a normal range and showing that Raleigh Radiology moved from 1.3 exams per hour to 1.43 exams per hour, yielding $140,000 in additional income. The same chart was used to determine the cost per exam using dollars on the y axis and time on the x axis to show that the cost per exam was reduced by $20, resulting in a savings of $122,000 per year. According to Dierolf, the Control Charts provide both average numbers, but also a range of normalcy. “That’s the real value of these control charts,” he said. “The idea is that there is natural variation in any process, in any measurement you are taking, and you don’t want to fight fires if you know statistically it will be in this range.” In conclusion, Dierolf left the audience with this advice: use what you have already, but schedule it better schedule more reduce the length of appointment times “It’s really important when you are going somewhere to know where you are starting from,” said Dierolf in his concluding remarks. “Money is the language of business, so if you can translate the impact of what you are doing into dollars, the business people understand dollars. We have a lot of analogies to the airline industry with perishable assets: an empty airline seat is gone.”