Secondary Logo

Journal Logo

Economics, Education, and Policy: Research Report

Rescheduling of Previously Cancelled Surgical Cases Does Not Increase Variability in Operating Room Workload When Cases Are Scheduled Based on Maximizing Efficiency of Use of Operating Room Time

Epstein, Richard H. MD, CPHIMS*; Dexter, Franklin MD, PhD

Author Information
doi: 10.1213/ANE.0b013e3182a0d9f6
  • Free


Even after controlling for the combination of surgical service and day of the week, there is substantial variability in operating room (OR) workload (i.e., total hours of scheduled cases plus turnovers) (Fig. 1).1 Intuitively, cancelling a case close to the scheduled day of surgery would contribute to the problem.2 However, in a recent study of decisions made by anesthesiologists and nurse managers in the OR scheduling office a few days before surgery, we tested this hypothesis, and came to the opposite conclusion: cancellations (slightly) reduce variability in services’ workloads among days, as measured in units of hours (P < 0.0001).3 This is not because many cancellations reduce the total hours close to zero resulting in no variability at all.3 Rather, cancellations soon before the day of surgery are associated with (slightly) increased net hours of cases scheduled.3

Figure 1
Figure 1:
Coefficients of variation of scheduled hours and turnover times of cases for different surgeons and for combinations of surgical service and day of the week. The data used to create the figure for each of the 74 studied surgeons were limited to days for which the surgeon performed at least 1 scheduled case. These surgeons were from 12 different services (e.g., cardiac or orthopedic surgery) and all had at least 64 days with at least 1 scheduled case. Among the 58 different combinations of these services and days of the week, there were 52 combinations satisfying that criterion and shown in the figure. The figure shows that most surgeons (70/74, P < 0.0001) have substantial (>30%) variability of the daily workload, as studied previously.18 , 19 The figure shows that most combinations of service and day of the week (43/52, P < 0.0001) have substantial (>30%) variability of the daily workload. Finally, the figure shows that a major reason for this variability is that individual surgeons encompassing the services have even larger variability in workloads (P < 0.0001, Kolmogorov-Smirnov 2-group test). The robust estimate for the coefficient of variation was calculated by multiplying a ratio by 1.28. The numerator of the ratio was the difference of the values at the 90th and the 10th percentiles. The denominator was the sum of the values at the 2 percentiles. The value of 1.28 equals the 90th percentile and −1 × the 10th percentile of the standard normal distribution. A robust estimate was used because the data are non-normally distributed with outliers. To create the dot plot, the calculated coefficients of variations were rounded to the nearest 3.33%.

This apparent paradox occurs because when cases are scheduled many weeks in advance, the workload for that date is known incompletely, since few cases have been scheduled and there may be many interim changes. When a cancelled case is replaced close to the day of surgery, the schedulers have a better idea of the anticipated workload.1,3–7 Consequently, schedulers can (and in practice do) fully fill but not exceed, the allocated hours for the specific OR into which the case is placed, allocated hours being statistically forecasted hours of the workday.1,3–6,8–12 The statistical forecast is made months before based on minimizing the weighted combination of expected hours of underutilized and overutilized hours of OR time.4,6,9,10 The total scheduled hours of cases of a surgical service 0 to 2 workdays before surgery is predicted by the difference between the statistically forecasted hours among its ORs for the workday and hours of cases scheduled so far (P < 0.0001).3

By no means does this result imply that cancellations on the workday before and on the day of surgery should be encouraged. In some countries, cases routinely are cancelled to prevent OR teams from working later than scheduled.13–18 However, such cancelations result in net increases in cost from physician, hospital, and societal perspectives (e.g., from the rework of getting the case rescheduled).14,15 Such cancellations result in patients and their families having taken off from work unnecessarily, cause them to sustain substantial waiting, and can cause patients to suffer adverse psychological consequences.19,20

There may be little strategic benefit to the anesthesiologists and hospital administrators investing resources to reduce their cancellation rates below typical (benchmarked) rates.3 If a relatively low incidence of cancellations does not cause increased variability in services’ workloads, this would be a useful finding when focusing strategic OR management initiatives.

However, the previous study considered only the effect on the schedule for the day the cancelled case originally was scheduled to be performed.3 The previous study did not consider the effect on the schedule for the subsequent day that the case was eventually performed. This article addresses the second part of case cancellation: how do “rescheduled” cases (cases previously cancelled on the day of surgery or on the working day immediately prior to the scheduled day of surgery) affect variability in workload on the day they are eventually performed.

The previous study was performed by service, not by surgeon.3Figure 1 shows the variability in workload by combination of service and day of the week (red dots) and by surgeon (blue dots). As can be seen, surgeons have even larger percentage variability in workloads among days with at least 1 scheduled case (Fig. 1).21 This arises in part because many surgeons perform few (e.g., 2–3) cases on their operative days.22

Suppose that every performed case was scheduled for the same duration and each day there were 0 or 1 rescheduled cases. Then, if rescheduled cases were increasing the variability in workload, the days with a rescheduled case would not be evenly distributed among days above or below the median workload. We could study the difference between 2 ratios,


. For the first ratio,

, we would consider only the days for which the scheduled workload of the surgeon was greater than the surgeon’s median workload. This ratio,

, equals the proportion of these days with a rescheduled case.

Figure 2 shows the variability in daily scheduled workload of performed cases of the most frequently operating surgeon with the most common variability in workload. The ratio

can be visualized by the number of dark blue dots divided by the number of dark blue or red dots. For the second ratio,

, we would consider only the days for which the scheduled workload of the surgeon was less than or equal to the surgeon’s median workload. This ratio,

, likewise equals the proportion of these days with a rescheduled case. In Figure 2, the ratio,

, can be visualized as the number of light blue dots divided by the number of light blue or green dots. We would study the difference between the ratios

, not the ratio of the ratios

, because the denominator can and occasional does equal zero causing an undefined result. If rescheduled cases were increasing the variability in workload (overutilized OR time), the difference between the ratios would exceed zero. In Figure 2, the difference would equal (9 dark blue dots/[9 dark blue dots + 12 red dots]) – (6 light blue dots/[6 light blue dots + 15 green dots]) = (9/21) – (6/21) = 14%. Emphasizing, a rescheduled case is a performed case that had previously been cancelled on the day of surgery or working day before the original date of surgery.

Figure 2
Figure 2:
Daily scheduled operating room workload of surgeon with the most common coefficient of variation. There were 19 surgeons with coefficients of variation between 52.1% and 54.5%, shown in Figure 1 by the many blue dots at 53% (Fig. 1). Among these 19 surgeons, 1 surgeon had the largest number of workdays with at least 1 case and that surgeon’s data was used in this figure. Among the 8 thirteen-week periods, the data shown in this figure are for the period for which that surgeon’s coefficient of variation equaled 53%. The dotted line shows the median scheduled workload of the surgeon, cases and turnovers. Consider the difference between 2 ratios
Figure 1
as described in the last paragraph of the preceding page.
Figure 1
= (9 dark blue dots/[9 dark blue dots + 12 red dots]).
Figure 1
= (6 light blue dots/[6 light blue dots + 15 green dots]). The dfference = (9/21) – (6/21) = 14%.

Although the preceding approach based on numbers of days is intuitive, it is inaccurate because it is suitable only if every case is scheduled for the same duration (e.g., 3 hours for mediastinoscopy, lung wedge resection, or esophagectomy). We recently performed a Bland–Altman analysis comparing cancellation rates calculated as a percentage of cases versus a percentage of scheduled durations and the results differed significantly both within and among services.23 Thus, we perform our study of performed cases in units of scheduled hours, not days. Doing so is not the same as performing an analysis in units of days while weighting each day by its scheduled hours, because the percentage contribution of cancelled cases and previously rescheduled cases to scheduled hours differs substantially among days.3

The primary end point of our study is the difference between 2 ratios with the numerators and denominators being scheduled OR hours,

. The first ratio,

, is the proportion of scheduled OR hours, including turnover times, attributable to rescheduled cases summed among the days with larger than median workloads. In Figure 3,

is visualized by the sum of the dark blue bars divided by the sum of the dark blue and red bars. The second ratio,

, is the same but among the days with a median or smaller workload. In Figure 3,

is visualized by the sum of the light blue bars divided by the sum of the light blue and green bars.

Figure 3
Figure 3:
Scheduled hours that contribute to the primary end point of the study represented graphically as calculated for 1 hypothetical surgeon, Dr. X. The dotted line shows the median scheduled workload of the surgeon, including both cases and turnovers. The displayed durations include the setup and cleanup times. The colors are the same as those of Figure 2. For example, on the first day, the surgeon can have performed one case scheduled for 4 hours and a second for 3 hours, the latter case rescheduled from a previous day. The calculated primary end point is unit less and equals the difference of 2 ratios, as described in the description of the Figure 3 in the beginning of the article. The primary study end point is
Figure 1
Figure 1
= (4 hours from dark blue bars/[4 hours dark blue bars + 28 hours red bars]). The
Figure 1
= (6 hours light blue bars/[6 hours light blue bars + 26 hours green bars]). The difference = (4/32) – (6/32) = −6%, less than zero.

Consider a hypothetical surgeon Dr. X who reschedules each patient whose case is cancelled just like any other patient waiting for surgery, regardless of the resulting days wait. Most of Dr. X’s cases have scheduled durations of either 3 hours or 4 hours, including setup and cleanup times. The net effect of cancellation would be that the surgeon’s percentage scheduled workload attributable to previously cancelled cases would be less on days when the surgeon has larger than median workload. For example, on Dr. X’s first day in Figure 3 with at least 1 case, the surgeon performed one case scheduled for 4 hours and a second case that had been rescheduled and was planned for 3 hours. Since this total equals the surgeon’s median of 7 hours, the 4 hours is shown in light green and the 3 hours in light blue. On the surgeon’s second day, the surgeon performed 2 cases, both scheduled for 4 hours. The sum of 8 hours exceeds the median of 7 hours. Because neither case was rescheduled, the 8 hours is displayed as a red bar. On the surgeon’s third day, the surgeon again performed two 4-hour cases, 1 of which was rescheduled. Because the sum of 8 hours exceeds the median, one 4 hours is shown in red and the other in dark blue. The value of the end point

in Figure 3 equals (4 hours from dark blue bars/[4 hours dark blue bars + 28 hours red bars]) – (6 hours light blue bars/[6 hours light blue bars + 26 hours green bars]) = (4/32) – (6/32) = −6%, less than zero. Because the denominator is the same, the end point is essentially the difference of the total rescheduled hours among all days that are busier versus not busier than the median.

Consider a different hypothetical surgeon, Dr. Y, who has scheduled OR days on Tuesdays and Thursdays and schedules each case in advance on the earliest date that the case can be performed without the total OR hours and turnovers exceeding 8.5 hours (Fig. 4). Nearly all of her cases are of one procedure, scheduled for 4.0 hours with 0.5 hours for the turnover. Cases are cancelled principally because the intensive care unit is full. When a case cancels, the surgeon has the case scheduled on her next OR day in an effort to increase patient satisfaction. On Dr. Y’s first day in Figure 4, the surgeon performed 2 cases each scheduled for 4 hours, plus there was 1 turnover time scheduled for 0.5 hours. Since this total equals the surgeon’s median of 8.5 hours, the 8.5 hours is shown in light green. On the surgeon’s second day, the surgeon performed 1 case, because a full intensive care unit caused her 2nd case to be cancelled. An add-on case was performed in her OR, but not performed by Dr. Y and so not shown in the Figure 4. On the third day, the surgeon performed 3 cases, with scheduled 13 hours, where 13 = 3 cases × 4 hours + 2 turnovers × 0.5 hours. There are 4.5 hours shown using a dark blue bar and 8.5 hours shown using a red bar. Each of the other days shown in the figure is the same as 1 of the first 3 days. The value of the end point

in Figure 4 equals (2 × 4.5 hours from dark blue bars/[2 × 4.5 hours dark blue bars + 2 × 8.5 hours red bars]) – (0 hours light blue bars/[0 hours light blue bars + {5 × 8.5 + 2 × 4.0 hours green bars}]) = (9/26) – (0/50.5) = 35%, greater than zero. The net effect would be that the rescheduled cases have increased the variability in OR workload for the surgeon. There would be more overutilized OR time.

Figure 4
Figure 4:
Scheduled hours that contribute to the primary end point of the study represented graphically as calculated for 1 hypothetical surgeon, Dr. Y. The dotted line shows the median scheduled workload of the surgeon, including her cases and turnovers. The colors are the same as those of Figures 2 and 3. The figure is described at the bottom of the first column of this page.

Finally, consider a third hypothetical surgeon Dr. Z, who has a nearly identical scheduled workload for each Monday and Friday she operates for each of the next 4 weeks. Her scheduled workload for most of these days equals her typical (median) workload. The few exceptional days have larger workloads than her median. Then, if a case scheduled for a different day were cancelled and added to one of the workdays with the median workload, that day would have a workload larger than her median. The net effect would be that the rescheduled cases have increased the variability in workload for the surgeon. Again, there would be more overutilized OR time.

The behavior involving case rescheduling likely differs among surgeons and thus study of overall activity as relevant to anesthesiologists and hospital staff needs to be performed with stratification by surgeon. We hypothesized that, in contrast to the described behavior of Drs. Z and Y, more commonly most hours of cases are rescheduled to be performed on days where the workload is less than or equal to the surgeon’s median (i.e., like Dr. X of Figure 3).


The study was approved by the Thomas Jefferson University IRB without requirement for written patient consent. Data from all cases scheduled to be performed between Sunday December 3, 2010 and Saturday November 30, 2012 were retrieved from the OR case scheduling system (ORSOS®, McKesson, San Francisco, CA). These data were case identifier, scheduled date of surgery, patient medical record number, scheduled case duration, primary surgeon/service, scheduled procedure code(s) for the case, actual OR location, time of OR entry (used to determine that the case was performed), and timestamps when cases were changed (scheduled, rescheduled, or cancelled).

The study was limited to cases performed in Thomas Jefferson University Hospital’s main surgical suites or adjacent ambulatory surgery center on nonholiday workdays. Anesthetics administered in other locations, such as the gastrointestinal endoscopy suites and the magnetic resonance imaging suite, were excluded. The 46,631 performed cases were identified by a date-time entry in ORSOS indicating that the patient actually entered the OR.

The 2238 cases that were not elective were excluded since by definition such cases cannot be rescheduled into the future. Cases were considered as nonelective if the patient’s American Society of Anesthesiologists Physical Status code contained an “E” modifier and the case was created within 5 hours of entering the OR. Physical status codes were retrieved by linking the ORSOS case identifier with the corresponding entry in the hospital’s anesthesia information management system (Innovian®, Dräger, Telford, PA).

Reconstruction of the OR schedules for every day as of 7:00 AM and 7:00 PM was accomplished by interrogating the audit tables and matching previously cancelled or rescheduled cases with those that were actually performed (Fig. 5). For each case performed, the patients’ medical record number was used to identify other (preceding) scheduled cases in the dataset. Cases for which the scheduled date of surgery was changed were considered cancelled at the timestamp for the change and rescheduled, as of that timestamp, to the new date. If a case was cancelled and a new case scheduled for a given patient for the exact same main procedure under a new case identifier, the first case identifier was replaced with the subsequent case identifier. There were 3361 such cases. If a case was cancelled and a new case scheduled for the same patient for a similar procedure (determined from manual inspection of the 2415 procedure codes used during the 2 years), the first case identifier also was replaced. There were 3080 such cases. For example, if the patient had been scheduled for thoracoscopic lung biopsy, that case was cancelled, and the patient was rebooked a week later to undergo open thoracotomy, the 2 cases were combined under the identifier of the performed case. If the same procedure by the same surgeon was cancelled, rescheduled, and performed, all on the same day, it was not considered to be a cancellation (i.e., functionally the schedulers were “uncancelling” the case). Because we were studying the impact of cancellations on performed cases, case scheduling events were attributed to the surgeon who did the procedure if another member of the surgeon’s service originally had scheduled the case.

Figure 5
Figure 5:
Time from cancellation until case is performed. The data match those used for the primary end point. The intervals displayed under the bars were calculated from the original date of surgery (for cases cancelled after 7:00 AM on the working day before surgery) until the case was subsequently performed. For the bars, the look-ahead period was limited to 1 year for all cases. For the dotted red line, the horizon ranged from 1 year to 2 years. If a case was rescheduled or cancelled several times after 7:00 AM on the working day before surgery, the date of surgery of the first such cancellation was used. There were 14% of cancellations that were not rescheduled within 1 year of the cancellation (148/1039 cases). There were 10% for which no rescheduling was observed over the 1 to 2 year range (103/1039).

The working day before surgery was determined for each workday based on the day of the week and holidays. For example, the working day prior to all Mondays and prior to the Tuesday after Labor Day was the preceding Friday. Cases were considered as having been cancelled after 7:00 AM or 7:00 PM on the working day before surgery if there was a timestamp for the case cancellation later than these times. For example, if a case scheduled for Wednesday was cancelled at 10:05AM on the preceding Tuesday, the case was counted as having been cancelled after 7:00 AM but not after 7:00 PM on the working day before surgery.

An unbiased Bayesian estimated duration was calculated for each case using the surgeon’s (scheduler’s) estimated duration and the historical data of cases of the same scheduled procedure code(s) and surgeon.,24–29,a The estimate was considered the scheduled duration of the case. From these, the scheduled workload was calculated for each surgeon and workday as the sum of the scheduled durations + (number of performed cases – 1) × mean turnover time. For example, if there were 3 scheduled cases and each was performed, there would have been 2 turnovers (between cases 1 and 2, and 2 and 3). That mean turnover time used for these workload determinations was 0.64 hours (38.6 minutes), calculated for actually performed cases in the dataset after excluding turnovers greater than 90 minutes.30

Data for each surgeon were obtained independently for 8 periods, each 13 weeks in duration, to allow for accurate calculations of the expected small percentages of cancelled cases.31 For example, the sum of the scheduled hours of cancelled cases was determined for each surgeon during each period. The mean ± standard error of the mean was calculated for each variable for each surgeon over the n = 8 periods. Workload data by period and surgeon are statistically independent among surgeons.18

The primary end point,

, was calculated for each surgeon for each of the n = 8 periods.b Surgeons were excluded if, for any period, either denominator in the 2 ratios was equal to zero (i.e., if the primary end point was undefined mathematically). Thus, each included surgeon performed at least 1 case on each of 2 days during each 13-week period. That minimum criterion was sufficiently small that, with a 5% cancellation rate, a surgeon may have no cancellations in some periods (i.e., numerator) to study. Thus, to be included, a surgeon also had to have at least 1 cancellation after 7:00 AM on the working day before surgery for at least half the periods. Using this criteria, we studied N = 74 surgeons, averaging 246 ± 13 elective cases per year on 86 ± 4 workdays. After the aforementioned exclusions, there were 37,318 performed cases suitable for analysis.

The primary end point,

, included a median threshold for the surgeon’s workload during each 13-week period and a less than or equal relationship (see the second page of the article). By design, it amounted to slightly more than half (51.8% ± 0.1%) of surgeons’ workdays with at least 1 case. Suppose that the hypothetical surgeon Dr. Z performs 1 long procedure on almost all days she operates. Her median scheduled workload would be the scheduled duration for that procedure. Days with workload less than or equal to the median would imply scheduling the cancelled case on a date that the surgeon had not yet scheduled a case. Most days with workload longer than the median would be scheduling 2 cases on the same day.

The overall managerial impact of surgeons and schedulers’ behavior on anesthesiologists and OR nursing is the pooled effect. From each surgeon there is a mean ± SEM calculated using n = 8 numbers, each from a long (13-week, i.e., statistically independent) period.25,30-32 DerSimonian and Laird random effect analysis was used to obtain the pooled mean estimate (I2 = 28%, χ2P = 0.014).33–38 Two-sided P-values were calculated using the Student t statistic with the sample size being the N = 74 surgeons.33


From 7:00 AM the working day before surgery through the day of surgery, 9.7% ± 0.6% of scheduled OR hours and 9.7% ± 0.5% of cases were cancelled. Among cases performed, 9.5% ± 0.5% of the scheduled hours and 9.5% ± 0.5% of the cases were previously cancelled (i.e., rescheduled to a later date and then performed). The 9.7% and 9.5% can be close because some cases were cancelled more than once. Among cases performed, 8.3% ± 0.5% of the scheduled hours and 8.4% ± 0.5% of the cases were previously cancelled precisely once.

Surgeons’ median workloads on days with at least 1 case were 8.3 ± 0.2 hours, close (P = 0.46) to the minimum shift duration of 8.5 hours, including a 0.5-hour lunch break.

The primary end point was a difference of ratios (see the second and third pages of the article). The percentage scheduled workload attributable to rescheduled cases was slightly less on days when the surgeon had larger than median workload (−0.7% ± 0.3%, P = 0.022). Only 1 surgeon may have performed significantly more hours of previously cancelled cases on days with larger than median workloads (all uncorrected P ≥ 0.04). If these N = 74 P-values were corrected for multiple comparisons, they would be larger (i.e., would support further that previously cancelled cases usually are not rescheduled into relatively busy days).


The results show that rarely did surgeons behave like the hypothetical surgeons Drs. Z and Y, whose rescheduling behavior increased their variability in workload (see the beginning of the article). There was overall a managerially irrelevant, but statistically significant (−0.7% ± 0.3%), trend for cancelled and rescheduled cases to be performed on a date during which the surgeon had a scheduled workload that was less than or equal to the surgeon’s median workload, including the previously cancelled case. Thus, rescheduling the cancelled cases convincingly did not overall increase variability in surgeons’ OR workloads. This finding is useful when combined with our recent finding that cancellation slightly reduces variability in OR workload on the date of cancellation, and paradoxically does so by increasing the net hours scheduled.3

Our findings can guide strategic decisions (e.g., by an OR executive committee) at hospitals such as the one studied with cancellation rates that are less than surveyed (benchmark) rates (see below). Investing resources to reduce the cancellation rate further may have no strategic advantage. Additional production-pressure not to cancel cases that currently are being cancelled for medical reasons may be counter-productive from a safety perspective.39 We recommend that these results be considered if cancellation rates are used in assessing anesthesiology group performance.40 We recommend also that cancellations not be interpreted as a system failure that increases variability in surgeons’ workloads.2 We recommend that anesthesiologists aim to reduce cancellation rates if above benchmarked averages,41 but otherwise focus on more strategically beneficial initiatives.

Although our results are limited to 1 U.S. hospital, we doubt that this is an important limitation for 3 reasons. First, we studied N = 74 surgeons’ behavior when scheduling cases many days before surgery. The surgeons’ behaviors were heterogeneous and not managed by a single office or enforced policies (“charter”).42 Second, suppose that there had been rigorous control (e.g., limiting scheduling behavior to avoid cases running past the end of allocated time), as described for some non-U.S. hospitals.11,12 Then, the surgeons would have been even less likely to scheduled cancelled cases on days that they were already busy (i.e., our results would be stronger in the direction observed). Third, our studied hospital’s incidence of cancellations was appropriately (for our study) modestly less than typical: 7.3% ± 0.4% of cases were cancelled after 7:00 PM the working day before surgery. That incidence compares with the German 12.2% for academic hospitals and 8.4% for large community hospitals.36 Matching the survey,36 our incidence was larger for General surgery (9.2% ± 1.0% cases, N = 25 surgeons) than Gynecology (2.0% ± 0.5% cases, N = 6).

The major limitation of our study is that likely the results are useful only for hospitals that use statistical forecasting of the durations of the workday for the ORs of each service on each day of the week (i.e., allocated time), appropriately calculated months in advance based on minimizing the inefficiency of use of OR time.1,3-5,10–12,15,43–48 These forecasts are used for staff scheduling months before surgery and then case scheduling starting early the workday before surgery.3–6,41 Our results are valid regardless of how scheduling was (is) done, but usefulness (i.e., conclusions) depends on the behavior previously studied for the day when the cancellation occurs.3 That behavior (see the beginning of the article) is highly sensitive to use49 of appropriate statistical forecasts because the forecasts include the probability of cancelled and add-on cases.3 We recommend strongly that hospitals where OR management decisions are not made based on economic grounds hesitate before applying our results. For example, suppose that, instead, a hospital treats the OR schedule as of 4:00 PM the workday before surgery as what the day of surgery will be. Then, of course, cancellations increase variability (predictive error from the schedule) because the schedule has changed. Anesthesiologists need economically (1) to assure that appropriate forecasts are calculated months in advance when staff scheduling is done and (2) to work with the scheduling office throughout the working day before surgery.1,3–6,9–12,38–44 In brief, good forecasts are made using historical data by service and day of the week. If future allocated time (i.e., the hours into which cases are scheduled) is too large, then there will be underutilized OR time.38 If allocated time is too small, there will be overutilized OR time.38 Based on the known relative cost of over- versus underutilized OR time, the optimal allocated hours to minimize the inefficiency of use of OR time can be calculated.1,10,38,40 Trained and untrained managers perform poorly at this optimization problem.44 The value of education is to increase leaders’ trust in the use of statistical methods and skill at evaluating when a recommendation may be based on incomplete information.50–53 The references include review articles9,10,52 and a brief review and sensitivity analysis for facilities in which most services have only 1 OR.40 Our course’s curriculum, lecture slides, and cases are available publically.46,47,c


Franklin Dexter is the Statistical Editor and Section Editor for Economics, Education, and Policy for the Journal. This manuscript was handled by Dr. Steven Shafer, Editor-in-Chief, and Dr. Dexter was not involved in any way with the editorial process or decision.


Name: Richard H. Epstein, MD, CPHIMS.

Contribution: This author helped design the study, conduct the study, and write the manuscript. This author is the archival author.

Attestation: Richard Epstein has approved the final manuscript.

Conflicts of Interest: Richard Epstein is President of Medical Data Applications, Ltd., whose CalculatOR™ software includes some of the analyses described in this article. The University of Iowa pays licensing fees to use the software for hospital consultations performed by its Division of Management Consulting.

Name: Franklin Dexter, MD, PhD.

Contribution: This author helped design the study, analyze the data, and write the manuscript.

Attestation: Franklin Dexter has approved the final manuscript.

Conflicts of Interest: The University of Iowa, Department of Anesthesia, Division of Management Consulting, performs some of the analyses described in this paper for hospitals. He has tenure and receives no funds personally, including honoraria, other than his salary and allowable expense reimbursements from the University of Iowa. Income from the Division’s consulting work is used to fund research.


a We used equations (1–4), (6), (13), and (16) of Ref. 24. Implementation is described in detail in Appendix 2 of Ref. 26. Bayesian methods are needed because there are many cases of rare combinations of scheduled procedure(s) and surgeon.25,28 The particular Bayesian method that we used is sufficient for unbiased estimation of the mean (expected value) of the duration of the case.26 Other Bayesian formulations likely have equal accuracy for this purpose.27,29 We chose the method used because it provides for estimates of the longest time a case might take and thus is in regular use at the studied hospital.24,26
Cited Here

b The analysis is not one of a risk difference. The units of the 2 denominators and 2 numerators are hours not counts (Figures 2-4). However, consider the reported demographic of the cancellation rate, a single ratio of counts. There are n = 8 such ratios for each surgeon, to be combined based on underlying normal distribution.31 The effective sample size is not the total number of cases the surgeon has scheduled, because the clustering of cancellations over time result in the violation of the binomial distribution assumption of independent events.31
Cited Here

c Accessed February 18, 2013.
Cited Here


1. He PB, Dexter F, Macario A, Zenios S. The timing of staffing decisions in hospital operating rooms: incorporating workload heterogeneity into the newsvendor problem. Manuf Serv Op. 2012;14:99–114
2. Tung A, Dexter F, Jakubczyk S, Glick DB. The limited value of sequencing cases based on their probability of cancellation. Anesth Analg. 2010;111:749–56
3. Dexter F, Shi P, Epstein RH. Descriptive study of case scheduling and cancellations within 1 week of the day of surgery. Anesth Analg. 2012;115:1188–95
4. Dexter F, Traub RD, Macario A. How to release allocated operating room time to increase efficiency: predicting which surgical service will have the most underutilized operating room time. Anesth Analg. 2003;96:507–12
5. Dexter F, Macario A. When to release allocated operating room time to increase operating room efficiency. Anesth Analg. 2004;98:758–62
6. Dexter F, Traub RD. How to schedule elective surgical cases into specific operating rooms to maximize the efficiency of use of operating room time. Anesth Analg. 2002;94:933–42
7. Dexter F, Epstein RH, Elgart RL, Ledolter J. Forecasting and perception of average and latest hours worked by on-call anesthesiologists. Anesth Analg. 2009;109:1246–52
8. Dexter F, Macario A, Traub RD. Which algorithm for scheduling add-on elective cases maximizes operating room utilization? Use of bin packing algorithms and fuzzy constraints in operating room management. Anesthesiology. 1999;91:1491–500
9. Dexter F, Epstein RH, Traub RD, Xiao Y. Making management decisions on the day of surgery based on operating room efficiency and patient waiting times. Anesthesiology. 2004;101:1444–53
10. McIntosh C, Dexter F, Epstein RH. The impact of service-specific staffing, case scheduling, turnovers, and first-case starts on anesthesia group and operating room productivity: a tutorial using data from an Australian hospital. Anesth Analg. 2006;103:1499–516
11. Van Houdenhoven M, van Oostrum JM, Hans EW, Wullink G, Kazemier G. Improving operating room efficiency by applying bin-packing and portfolio techniques to surgical case scheduling. Anesth Analg. 2007;105:707–14
12. Hans E, Wullink G, van Houdenhoven M, Kazemier G. Robust surgery loading. Eur J Oper Res. 2008;185:1038–50
13. Ibrahim GM, Tymianski M, Bernstein M. Priority setting in neurosurgery as exemplified by an everyday challenge. Can J Neurol Sci. 2013;40:378–83
14. Tessler MJ, Kleiman SJ, Huberman MM. A “zero tolerance for overtime” increases surgical per case costs. Can J Anaesth. 1997;44:1036–41
15. Stepaniak PS, Mannaerts GH, de Quelerij M, de Vries G. The effect of the operating room coordinator’s risk appreciation on operating room efficiency. Anesth Analg. 2009;108:1249–56
16. Chiu CH, Lee A, Chui PT. Cancellation of elective operations on the day of intended surgery in a Hong Kong hospital: point prevalence and reasons. Hong Kong Med J. 2012;18:5–10
17. McIntosh B, Cookson G, Jones S. Cancelled surgeries and payment by results in the English National Health Service. J Health Serv Res Policy. 2012;17:79–86
18. Hovlid E, Bukve O, Haug K, Aslaksen AB, von Plessen C. A new pathway for elective surgery to reduce cancellation rates. BMC Health Serv Res. 2012;12:154
19. Tait AR, Voepel-Lewis T, Munro HM, Gutstein HB, Reynolds PI. Cancellation of pediatric outpatient surgery: economic and emotional implications for patients and their families. J Clin Anesth. 1997;9:213–9
20. Ivarsson B, Kimblad PO, Sjöberg T, Larsson S. Patient reactions to cancelled or postponed heart operations. J Nurs Manag. 2002;10:75–81
21. Dexter F, Masursky D, Ledolter J, Wachtel RE, Smallman B. Monitoring changes in individual surgeon’s workloads using anesthesia data. Can J Anaesth. 2012;59:571–7
22. Dexter F, Macario A, Traub RD, Lubarsky DA. Operating room utilization alone is not an accurate metric for the allocation of operating room block time to individual surgeons with low caseloads. Anesthesiology. 2003;98:1243–9
23. Ehrenfeld JM, Dexter F, Rothman BS, Johnson AM, Epstein RH. Case cancellation rates measured by services differ if based on the number of cases or the number of minutes cancelled. Anesth Analg. 2013;117:711–6
24. Dexter F, Ledolter J. Bayesian prediction bounds and comparisons of operating room times even for procedures with few or no historic data. Anesthesiology. 2005;103:1259–167
25. Dexter F, Macario A, Ledolter J. Identification of systematic underestimation (bias) of case durations during case scheduling would not markedly reduce overutilized operating room time. J Clin Anesth. 2007;19:198–203
26. Dexter F, Epstein RH, Lee JD, Ledolter J. Automatic updating of times remaining in surgical cases using bayesian analysis of historical case duration data and “instant messaging” updates from anesthesia providers. Anesth Analg. 2009;108:929–40
27. Stepaniak PS, Heij C, Mannaerts GH, de Quelerij M, de Vries G. Modeling procedure and surgical times for current procedural terminology-anesthesia-surgeon combinations and evaluation in terms of case-duration prediction and operating room efficiency: a multicenter study. Anesth Analg. 2009;109:1232–45
28. Dexter F, Dexter EU, Ledolter J. Influence of procedure classification on process variability and parameter uncertainty of surgical case durations. Anesth Analg. 2010;110:1155–63
29. Eijkemans MJ, van Houdenhoven M, Nguyen T, Boersma E, Steyerberg EW, Kazemier G. Predicting the unpredictable: a new prediction model for operating room times using individual characteristics and the surgeon’s estimate. Anesthesiology. 2010;112:41–9
30. Dexter F, Epstein RH, Marcon E, Ledolter J. Estimating the incidence of prolonged turnover times and delays by time of day. Anesthesiology. 2005;102:1242–8
31. Dexter F, Marcon E, Epstein RH, Ledolter J. Validation of statistical methods to compare cancellation rates on the day of surgery. Anesth Analg. 2005;101:465–73
32. Dexter F, Macario A, Qian F, Traub RD. Forecasting surgical groups’ total hours of elective cases for allocation of block time: application of time series analysis to operating room management. Anesthesiology. 1999;91:1501–8
33. DerSimonian R, Laird N. Meta-analysis in clinical trials. Control Clin Trials. 1986;7:177–88
34. Whitehead A Meta-analysis of controlled clinical trials. 2002 West Sussex, England John Wiley & Sons, Ltd:50, 59, 60, 82, 83, 89, 91–6
35. Higgins JP, Thompson SG. Quantifying heterogeneity in a meta-analysis. Stat Med. 2002;21:1539–58
36. Sidik K, Jonkman JN. A comparison of heterogeneity variance estimators in combining results of studies. Stat Med. 2007;26:1964–81
37. Jackson D, Bowden J, Baker R. How does the DerSimonian and Laird procedure for random effects meta-analysis compare with its more efficient but harder to compute counterparts? J Stat Plan Inference. 2010;140:961–70
38. Ledolter J, Dexter F. Analysis of interventions influencing or reducing patient waiting while stratifying by surgical procedure. Anesth Analg. 2011;112:950–7
39. Gaba DM, Howard SK, Jump B. Production pressure in the work environment. California anesthesiologists’ attitudes and experiences. Anesthesiology. 1994;81:488–500
40. Stepaniak PS, Dexter F. Monitoring anesthesiologists’ and anesthesiology departments’ managerial performance. Anesth Analg. 2013;116:1198–200
41. Schuster M, Neumann C, Neumann K, Braun J, Geldner G, Martin J, Spies C, Bauer MCASCAES Study Group. . The effect of hospital size and surgical service on case cancellation in elective surgery: results from a prospective multicenter study. Anesth Analg. 2011;113:578–85
42. Ernst C, Szczesny A, Soderstrom N, Siegmund F, Schleppers A. Success of commonly used operating room management tools in reducing tardiness of first case of the day starts: evidence from German hospitals. Anesth Analg. 2012;115:671–7
43. Strum DP, Vargas LG, May JH, Bashein G. Surgical suite utilization and capacity planning: a minimal cost analysis model. J Med Syst. 1997;21:309–22
44. Dexter F, Epstein RH, Marsh HM. A statistical analysis of weekday operating room anesthesia group staffing costs at nine independently managed surgical suites. Anesth Analg. 2001;92:1493–8
45. Pandit JJ, Dexter F. Lack of sensitivity of staffing for 8-hour sessions to standard deviation in daily actual hours of operating room time used for surgeons with long queues. Anesth Analg. 2009;108:1910–5
46. Dexter F, Wachtel RE, Epstein RH, Ledolter J, Todd MM. Analysis of operating room allocations to optimize scheduling of specialty rotations for anesthesia trainees. Anesth Analg. 2010;111:520–4
47. Sulecki L, Dexter F, Zura A, Saager L, Epstein RH. Lack of value of scheduling processes to move cases from a heavily used main campus to other facilities within a health care system. Anesth Analg. 2012;115:395–401
48. Wang J, Dexter F, Yang K. A behavioral study of daily mean turnover times and first case of the day start tardiness. Anesth Analg. 2013;116:1333–41
49. Wachtel RE, Dexter F. Review article: review of behavioral operations experimental studies of newsvendor problems for operating room management. Anesth Analg. 2010;110:1698–710
50. Dexter F, Willemsen-Dunlap A, Lee JD. Operating room managerial decision-making on the day of surgery with and without computer recommendations and status displays. Anesth Analg. 2007;105:419–29
51. Wachtel RE, Dexter F. Curriculum providing cognitive knowledge and problem-solving skills for anesthesia systems-based practice. J Grad Med Educ. 2010;2:624–32
52. Dexter F, Masursky D, Wachtel RE, Nussmeier NA. Application of an online reference for reviewing basic statistical principles of operating room management. J Stat Educ. 2010;18(3)
53. Prahl A, Dexter F, Braun MT, Van Swol L. Review of experimental studies in social psychology of small groups when an optimal choice exists and application to operating room management decision-making. Anesth Analg. (in press)
© 2013 International Anesthesia Research Society