Maximizing Operating Room Utilization

Before managed care and capitated payments, scheduling surgical [1] cases was a simple matter. Essentially, all available operating rooms (ORs) were staffed 8 h a day, and procedures were booked on a first-come, first-served basis. Hospital collections for OR care often exceeded five times hospital costs, so there was little concern over how efficiently the ORs were scheduled. The primary goal was to provide as much care and surgical convenience as possible. As available ORs became filled with cases and surgeons experienced difficulty in booking elective cases, blocks of OR time were given to busy surgeons or surgical services. Although use of block time tended to reduce utilization of the ORs, during the "golden era" of fee-for-service reimbursement, utilization rates as low as 20% still produced a positive bottom line. Because rewarding block time greatly increased surgeon satisfaction, hospital administrators felt pressure to grant it, especially if competing hospitals provided it. For almost all hospitals, recent changes in health-care have produced two trends that have dramatically increased the concern over the efficiency of OR use. First, reimbursement for OR care has been decreased because of Diagnosis Related Group payments, reduced fee-for-service rates, or capitated payment plans. Second, predicted reductions in the amount of surgery have not materialized [1]. Although individual surgical practices may have ended due to group or insurance carrier changes, the number of surgeries being performed today has actually increased over the decade. Thus, hospitals must accommodate surgical practices that are busy and growing yet do not produce the revenue stream that supports low rates of OR utilization. How, then, should OR scheduling be performed to maximize OR use? Most centers have traditionally addressed this problem by awarding block time, then encouraging its maximal use. Administrative efforts focused on trying to decide which services will be granted block time and what utilization rate should be required to maintain it. The former is most often historically determined (i.e., which services were granted it first), whereas the latter is often fruitlessly debated. Unfortunately, before the work by Dexter et. al. [2], which is presented in this issue, there had been no studies to determine which scheduling strategy for block time maximizes OR use [3]. Thus, decisions were based on empiric beliefs and, frequently, political issues. Mounting pressure to reduce the costs of operative care has changed the priorities for such decisions. In their study, Dexter et al. used computer simulation to address the many variables that can be considered when attempting to optimize surgical scheduling. The variables examined included the average case duration, average patient waiting time for surgery, number of hours in each block, number of blocks per surgeon, and scheduling strategy. For the latter, the authors evaluated the various possibilities of what is called "bin-packing", in which surgeries are the items to be packed into the bins of surgical scheduling blocks. The end point for all the simulations was the same: average OR utilization. This was correctly defined as the total amount of time patients are in the OR, plus turnover times, divided by available block time [4]. Simulating clinical practice in this manner had tremendous advantages: it allowed the authors to test changes in all of the variables without affecting actual care delivery. It permitted multiple changes in inputs that would have required many years and thousands of patient interactions to duplicate in real life. Further, simulation allowed testing of the sensitivity of OR utilization to changes in one variable; that is, it allowed measurement of the percent change in OR utilization for a clinically relevant change in one of the variables. Many of the changes the authors measured could have been qualitatively predicted. However, the authors' study provides the advantage of quantifying the extent of these changes. What were their findings? Before performing the computer simulation, Dexter et al. surveyed patients to determine what they considered to be the longest acceptable wait time for scheduling elective surgery; within a 95% confidence limit, the median value was 2 wk. There was little difference in OR use for each of the four possible scheduling algorithms. Generally, full-day blocks (7 or 8 h) produce greater OR use than half-day blocks, but no significant improvement in OR use can be obtained by changing from a 7-h to an 8-h block duration. If the wait time is 1 wk and average case length was >or=to2 h, then two 4-h blocks were used more efficiently than one 8-h block. If the wait time is 2 wk, however, there is little advantage of two half-day blocks over full-day blocks. Variability in surgical practice causes widely varying rates of OR use, even if a surgeon or service is always attempting to fill their block time. However, because a given surgeon's utilization tends to stay the same over time (auto-correlation), adjusting block time allocations based on short-term historical data (weeks or months) is not likely to improve OR utilization. Similarly, auto-correlation of the wait time limits the usefulness of adjusting block times based on successively measured wait times for a single surgeon. The variable that has the greatest effect on OR utilization is the average length of time that patients have to wait to undergo surgery. Depending on how the other variables are set, increasing wait time from 1 to 2 wk increased OR utilization from 6% to 29% (median 13%), whereas an increase from 2 to 3 wk resulted in a 0% to 12% (median 5%) improvement. To put this in perspective, the authors point out that a 10% improvement in OR utilization would allow closing of three ORs in a 30 OR suite without decreasing the amount of surgery performed. Analyzed from a cost standpoint, the savings in OR personnel costs that would accrue from a 10% improvement in utilization would be the equivalent of reducing the costs of all of the anesthetic drugs to zero! Would these simulated results be produced if the authors' suggestions were followed in the real world? To those who have spent hours at OR committees arguing over block time allocations, it is appealing to accept the conclusion that changes in individual surgeon block times or wait times have little effect on overall utilization. As unfilled blocks are generally used to accommodate add-on cases or cases from surgeons without block time, it certainly seems reasonable to adjust block times based on factors other than historical averages. The typical OR suite has so many alterations in the daily schedule, because of cancellations, add-ons, and emergencies, that following the authors' suggestion to use the simple "next fit" algorithm also makes sense. Likewise, fine tuning of OR use by making small adjustments in block time allotments does not seem worth the effort. The large gains in OR utilization that can be realized (in theory) by increasing wait time may be dependent on the number of ORs being scheduled in the real world. With 30 ORs from which to choose, it may be easy to schedule them such that a chosen average and maximal wait time is achieved. If only 10 ORs are available, significant efficiency improvements may produce wait times longer than is acceptable. In the authors' model, cost-savings accrue with increased OR utilization because "down" time and overtime are minimized. This assumes, however, that all OR personnel work a minimal (e.g., 8 h) shift and are paid at higher hourly rates past their base shift. If the duration of shifts were adjusted to match surgical need, however, then the positive economic effects of scheduling improvements are reduced. Further, improvements in OR use do not decrease the cost of salaried or fee-for-service personnel (although it certainly would shorten their hours in the OR). Of course, by focusing on only one output-utilization of OR block time-the authors did not consider other goals of an OR scheduling strategy. These include surgeon convenience, availability of ORs for urgent and emergent cases, and availability of open time for surgeons who do not have block time. It would be interesting to see whether computer simulation could predict the optimal balance of open versus block time. Perhaps the most contentious issue to address may be how to implement the optimization strategies. As noted by the authors, minimizing OR costs while maintaining acceptable patient wait times requires moving control of the actual surgical data from the surgeon or patient to the OR scheduler. The financial savings that would accrue from such a move seem to be worth the investment in time and effort required to convince surgeons to relinquish such control. Dexter et al. are to be congratulated for providing the first comprehensive analysis to support a rational, centralized scheduling policy.