In this study, we highlight the significant variability in efficiency across NORA locations using performance frontiers. While some NORA sites achieve efficiency similar to that of traditional OR locations, others show substantial differences, underscoring the need for targeted interventions. At our institution, NORA block times are dictated by institutional practices rather than data-based analysis of demand and efficiency. Mcintosh et al. demonstrated that appropriate block allocations are essential, and our approach enables leaders to develop site-specific strategies based on each site’s unique challenges [18]. In so doing, performance frontiers can be leveraged for strategic development and tactical decision-making.

Further, this study explores the inherent constraints and pressures imposed by these predetermined hours, revealing potential inefficiencies that arise from a lack of flexibility in operational planning. While each NORA location may request (or require) dedicated anesthesia services, multiple services are typically competing for a finite set of anesthesia resources. This constraint may limit how much operational efficiency may be improved through workflow measures used in the OR, which may negatively impact anesthesia groups and hospitals alike [7, 9]. As such, increasing NORA volumes might be a growing impediment to the financial health of both academic and private anesthesia groups. Though several strategies have been proposed to address efficiency improvement ranging from hands-off cost estimates for anesthesia subsidy negotiations to active day-before decision making, heretofore, none have shifted their viewpoint to consider the anesthesia provider as the constrained resource [19, 20].

With performance frontiers, we have demonstrated that several NORA locations at our institution have performance frontiers comparable to the main and cardiac ORs, including advanced gastroenterology (GI), electrophysiology (EP), and interventional radiology (IR) (Fig. 3). Advanced GI and EP are displaced upward and leftward along the frontier, indicating that they trend toward under-utilized time. Additional case volume (perhaps through addition of another proceduralist) or reduced anesthesia resources (e.g., less scheduled block time or an anesthesia resource shared with other NORA locations) would shift their points downward and rightward. On the other hand, IR demonstrates a high degree of over-utilized minutes. This indicates that the service might require additional anesthesia resources or that there is an inherent variability of the workload (e.g., after-hours cases). Tactically, the anesthesiology department could extend their scheduled block time or provide another block allocation during daytime work hours, recognizing that there might be a net loss of operational efficiency for anesthesia providers.

In contrast, inefficient, sporadically utilized locations consistently demonstrate both under- and over-utilized minutes, as shown in Fig. 4. Similarly, Schottel and colleagues showed that trauma services experience significant variability in workload and consistently face inefficiencies with both under- and over-utilized minutes [17]. This inefficiency underscores the challenges of managing capacity-based services using traditional metrics. Neurointerventional radiology is the archetypal example at our institution, given that emergent stroke thrombectomies occur at any time of day and nearly always require anesthesia services. This sporadic resource utilization is reflected graphically in Fig. 4, where its performance frontier curve is much farther from the origin than those of more efficient sites in Fig. 3. At an operational level, most clinical directors simply reconfigure the staffing available to provide anesthesia coverage on an ad hoc basis.

Implementing a shared “sandbox” approach allows for flexibility and adaptation based on demand and operational dynamics, optimizing real-time resource allocation. A flexible, shared scheduling system has been previously shown to improve utilization of both elective time-in-block as well as opportunity-unused time, accommodating increased case volumes without additional resources [21]. Varied operating hours stem from several factors, such as patient volume, equipment considerations, location oversight philosophy, lack of block time, limited interoperability due to room/equipment specialization, emergency procedures, and staffing patterns. A dynamic resource allocation model better addresses the challenges of capacity-based services by shifting from rigid block allocations to a system that reallocates resources based on real-time demand. Such on-demand staffing necessarily requires a certain amount of slack in the system for emergencies, but anesthesia departments should not be left subsidizing this inherent inefficiency. Presumably, performance frontiers can justify consolidation of resources, the expansion of anesthesia and nursing coverage, or even the institutional support necessary for structural investments.

As a final example, an ambulatory center-like cluster is also noted hugging the y-axis in Figs. 1 and 2, demonstrating significant amounts of under-utilized time, consistent with previously reported data [15]. Ambulatory surgical centers mostly cover elective, outpatient surgeries, and the variability in both case length and hours is less variable than mixed patient settings. Here, the hospital administrators can increase case volume, reduce the number of rooms (e.g., from 6 to 4), or shorten block allocations. For anesthesia services, staff scheduling considerations should include concurrencies, case complexity, regional anesthesia demands, and post-operative care unit coverage.

While performance frontiers highlight whether a site maps toward one end of the spectrum, they do not diagnose the root causes behind these patterns (e.g., sterile processing delays, turnover bottlenecks, or other factors that may initially seem unrelated to anesthesia staffing). However, they can be used to quickly pinpoint these misalignments and guide targeted interventions, even when the root causes themselves lie outside the direct control of anesthesia services. For example, if under-utilized time is driven by sterile processing delays and over-utilized time is also present, resolving the equipment delays would reduce both inefficiencies, shifting data points toward the origin (0,0). Alternatively, anesthesia groups can adapt their staffing models, decoupling their schedule from procedural block time, flexing in and out of rooms “on demand” rather than providing continuous coverage. This staffing model would minimize idle time while still meeting demand, thereby moving the data points closer to the origin. If under-utilized time is driven by delays, but over-utilized time is minimal, fixing those delays may shift under-utilized time to the end of the workday—highlighting a new opportunity to reduce block time, adjust staffing, or add additional surgical cases. These insights illustrate how performance frontier data can be used iteratively to address broader system issues and refine local strategies for resource allocation and operational improvement. Table 1 illustrates how performance frontier data can inform context-specific interventions; examples shown are hypothetical and not intended to be comprehensive or prescriptive.

Although no formal operational changes have been implemented in response to our performance frontier analysis, its development coincided with a request for expanded anesthesia support by the interventional radiology service. Informed in part by insights from our analysis, departmental leaders piloted a second anesthesia-covered IR room one day per week (Mondays). This provided a natural opportunity to apply the framework retrospectively. Using anesthesia billing data from non-holiday weekdays between 1 November 2024 and 30 April 2025, we compared frontier curves for Mondays versus non-Mondays, again assuming 10-hour primetime blocks, normalized by number of days. The Monday PF curve was significantly closer to the origin, reflecting significant reductions in both under-utilized (p < 0.01) and over-utilized (p = 0.02) time (Fig. 5). Case volume data for the period (125 cases on Mondays, averaging 4.8 cases per day, vs. 362 cases on non-Mondays, averaging 2.8 cases per day) suggest that while some volume was new, much was likely redistributed from end-of-day overages into added prime-time capacity. While unmeasured differences between Monday and non-Monday scheduling may limit interpretation, this result retrospectively supports the performance frontier-informed strategy outlined in Table 1 and demonstrates improved alignment between anesthesia resources and procedural demand.

Performance frontier curves for interventional radiology (IR) cases on Mondays versus non-Mondays following the addition of a second anesthesia-covered IR room one day per week

There are several limitations to the present analysis. First, all data comes from a single academic medical center’s primary hospital and the results may not be generalizable to other locations. Several authors have identified significant productivity differences between and within anesthesia groups across academic and private practice settings [10, 13]. The specific frontier shapes presented here should not be interpreted as benchmarks, but rather as exemplars of the methodology. Sites with highly irregular financing models or extremely low case volume (e.g., fewer than 1 case per day) were excluded from analysis, which may further limit generalizability. Second, the present study does not distinguish any difference in cost between under- and over-utilized time. Since portions of fixed costs for “prime” time teams may differ from those covering end of day and overnight work, strategies to improving a given site’s operational efficiency must balance the clinical workload and costs.

While it may be enticing to develop cross-institutional benchmarking of NORA performance frontiers, defining a universal frontier is currently impractical. Doing so would require industry-wide definitions and implementations of block time, as well as adjustments for local case-mix, payor mix, staffing ratios, and regional labor/overhead costs, all of which vary markedly among hospitals. The primary strength of this approach therefore lies in intra-institutional use, allowing each site to function as its own control while also permitting rapid, locally tailored iteration of staffing or scheduling changes. Future studies incorporating direct costs and revenues may facilitate more meaningful inter-institutional comparisons. While performance frontiers cannot overcome broader system-level inefficiencies, they nonetheless provide a pragmatic, site-specific framework to guide incremental improvements in resource allocation within each institution’s unique environment.

Our study utilized 7:00 AM until 5:00 PM on weekdays as “NORA block time” and block allocations may vary across institutions. Not only do these hours not represent typical anesthesiologist working hours nationwide, but they may also not represent the typical operating hours of certain NORA sites. At our institution, these NORA location hours are effectively set by hospital policy, aligning with nursing shifts rather than data-driven optimization. When expanding anesthesia coverage to new NORA service lines, we have previously established full-day coverage up front, anticipating that case volumes will quickly grow to match capacity. However, this anticipated growth may not materialize, resulting in coverage models that exceed actual clinical need. Such fixed scheduling impacts resource utilization, underscoring the need for performance frontiers to periodically evaluate and recalibrate allocation. This adaptable methodology can similarly be tailored to meet the needs of other individual sites or institutions as well.

For this analysis, we chose to use anesthesia billing time as our measure of resource utilization rather than procedure time. While procedure time uses the proceduralist as the constrained resource and room occupancy time centers on the room itself, using anesthesia billing time highlights the anesthesia provider as the limiting factor. This approach not only captures the full scope of provider utilization—encompassing all pre- and post-procedure patient-related activities—but also allows us to adjust based on actual demand. Unlike proceduralist or room availability, which are beyond our control, the staffing of anesthesia providers can be adjusted to align with trends in utilization. Our approach assumes that the most efficient systems will plan and conduct their operative work during allocated times and conduct no work during non-allocated times. Despite these potential shortcomings, our study effectively demonstrates the utility of applying performance frontier methodology to the NORA setting, providing a baseline against which to evaluate subsequent successes or failures.

In closing, clinical operations teams can avoid the difficulties of comparing productivity across different sites or against national standards simply by using their own service as the comparator. Future studies employing performance frontiers can extrapolate beyond the current metrics, further highlighting the gap between the present state and aspirational benchmarks. By facilitating strategic planning and continuous assessment of tactical and operational decisions, perioperative services can leverage performance frontiers to optimize operations, improve efficiency and procedural timeliness, and perhaps, increase profitability.