Secondary Logo

Journal Logo

Uncertainty in Knowing the Operating Rooms in Which Cases Were Performed Has Little Effect on Operating Room Allocations or Efficiency

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

doi: 10.1097/00000539-200212000-00048
TECHNOLOGY, COMPUTING, AND SIMULATION: Research Report
Free

At many US surgical facilities, applying the previously published method that maximizes the efficiency of use of operating room (OR) time is an effective way to optimize the allocation of OR time. Results are resistant to small errors in recorded OR times. However, at some facilities, the OR information systems data have as much as a 10% error in the correct OR where each case took place. This decreases the total OR time attributed to each service, which is the basis for the allocation method. Such errors could result in incorrect OR allocations and increased OR staffing costs. Expensive and time-consuming data-cleaning steps may be required to resolve the actual OR allocation for each case. We used 1 yr of data from a large, tertiary academic hospital to investigate, through simulation, how increasing levels of error in the correct OR affect OR efficiency and allocations. To apply noise to the data, the actual ORs were changed randomly to unique, “unknown” rooms. At a 30% error level, OR allocations decreased by 4.8%, and costs increased by 1.4% relative to knowing the actual location of every case. Only 1 of 11 surgical services had an allocation decrease at room error rates of less than 25%. We conclude that, in most circumstances, data-cleaning steps to resolve uncertainty in OR locations are not necessary to make accurate OR allocations.

*Department of Anesthesiology, Jefferson Medical College, Philadelphia, Pennsylvania, and MDA Ltd., Jenkintown, Pennsylvania; and †Division of Management Consulting, Department of Anesthesia, University of Iowa, Iowa City, Iowa

MDA Ltd. developed and distributes the CalculatOR™ software that performs some of the analyses described in this article.

August 7, 2002.

Address correspondence to Franklin Dexter, MD, PhD, Division of Management Consulting, Department of Anesthesia, University of Iowa, Iowa City, IA 52242. Address e-mail to Franklin-Dexter@ UIowa.edu.

At most hospitals and freestanding surgical centers in the United States, surgeons and patients choose the day of surgery, and elective cases are not turned away. At such facilities, the most effective way to increase the efficiency of use of operating room (OR) time is to allocate this OR time appropriately to surgical services (1–9). This allocation approach uses each service’s historical total hours of cases, including turnover times, as recorded in an OR information system (ORIS). The method considers turnover time as the interval from when a patient exits the OR until the next patient enters. For each service, OR efficiency is maximized by minimizing the sum of the cost of under-utilized OR time (regular staffed hours during which no cases were performed) and the cost of over-utilized OR time (hours of cases performed after the end of the regular workday, or “overtime”) (1–3). This approach, and this article, does not apply to facilities in which the total hours of surgery each day are fixed (i.e., where elective cases are declined if they cannot be completed by the end of regularly scheduled hours).

Most facilities have some form of an ORIS in place, with data entered either by using computers located in each OR or, after surgery, by clerks using paper logs that were transcribed in the OR. However, few facilities currently use the information collected in their ORIS to optimize surgical service OR allocations on the basis of maximizing OR efficiency. The science to perform such optimization has only recently been developed (1–3,8), and the methodology to actually perform the analyses each month and then implement the recommendations is slowly being learned by organizations.

We have observed that when facilities first begin to use their ORIS to make such management decisions, they often discover errors in the data. Most often, errors are identified by there being two cases recorded as overlapping in the same OR. This can occur if a case is moved from its scheduled OR to another OR, but this information is not updated in the ORIS. This data error can occur if the wrong OR is entered into the ORIS. It can be caused by recording incorrect OR entrance and/or exit times. Also, it can be caused by switching am and pm in recorded times (e.g., listing the case as having been performed from 2:00 am to 4:00 am instead of 2:00 pm to 4:00 pm). We refer to this uncertainty in knowing the actual OR in the ORIS data as “room noise.” The practical issue addressed by this study was to determine the extent to which uncertainty in knowing the actual ORs in which cases were performed (i.e., room noise) affects the performance of the OR allocation method (1–3).

There are two methods that we know of to deal with room noise. One is to install a new or modify an existing ORIS and collect new data. Depending on the ORIS in use and the implementation features selected, some, but usually not all, of these types of errors can be prevented. Once a new ORIS has been installed, only a month and a half of new data are adequate to improve OR efficiency (8). However, in practice, hospitals usually prefer to figure out a way to use their existing data. The other approach in dealing with overlapping cases is to change the recorded OR of each case that overlaps into a unique “unknown” OR (1). This process decreases the total hours of cases, including turnover times, attributed to the surgical service doing the cases, because it decreases the total turnover time. Although, in theory, one could infer the true OR in which the misspecified case was performed, this process is manual, time-consuming, expensive, and often arbitrary. Knowing the sensitivity of the OR allocation method to room noise would provide guidance as to what steps, if any, need to be taken to eliminate room noise from ORIS data when making OR allocations to maximize OR time efficiency.

Back to Top | Article Outline

Methods

We analyzed 1 yr of data from a large Midwestern tertiary academic hospital where the actual OR was known for every case. Baseline OR allocations and the total cost were determined for each combination of the day of the week and service by using the previously described method to maximize OR efficiency (1–3). Parameter values chosen for these calculations were 1) a relative cost of over-utilized to under-utilized OR time of 1.75 and 2) a regularly scheduled workday of 8 h. Briefly, the expected efficiency of use of OR time (defined previously) was obtained for each day of the week and service by using 0 allocated hours (0 ORs), 8 allocated hours (1 OR), 16 allocated hours (2 ORs), and so forth. The optimal OR allocation was the number of 8-h ORs yielding the highest expected efficiency of use of OR time (1,2). For each day of the week, the services with no allocated OR time were assigned to a catchall service that we refer to as the OTHER service. The optimal allocation for the OTHER service was then calculated (1).

The OR allocations were calculated for every service on each day of the week by using the historical workload of the service on that day of the week, measured by its total hours of cases, including turnover times. To account for the possible presence of scheduled delays between contiguous cases in a given OR, a maximum turnover time of 60 min was used if the calculated turnover exceeded this value.

Room noise was injected into the data by randomly changing the recorded ORs of cases in the baseline dataset to unique “unknown” ORs. Only the OR was changed, in that all cases were included in the analysis, cases were attributed to the original service, and no case durations were changed. The number of cases for which the OR was changed to unknown equaled the percentage room noise multiplied by the number of cases in the dataset, divided by 100. Once room noise was added, the total hours of elective cases, including turnover times, were recalculated for each service and day of the week combination. The OR allocations were determined by using the allocation method to maximize OR efficiency (1–3). “Staffing costs” (i.e., labor costs) were estimated by adding the number of hours of allocated OR time plus 1.75 multiplied by the hours of over-utilized OR time. The value of 1.75 represents time and a half for overtime plus an incremental factor of 0.25, which represents indirect costs, such as job dissatisfaction associated with having to stay late on call to complete cases (1).

Fifty simulations were run at each room noise value between 5% and 30%, in 5% increments. Although our experience is that ORIS datasets typically have 0% to 10% room noise, we applied up to 30% noise to ensure that we could detect an important effect of room noise, if it were present. At each level of room noise, the mean was determined for the percentage difference in cost relative to the baseline cost and for the percentage difference in the total hours of allocated OR time relative to its baseline value. Corresponding 95% confidence intervals (CIs) for the true mean were calculated by using Student’s t-distribution.

Back to Top | Article Outline

Results

There were 12,264 cases in the dataset assigned to 11 surgical services. Nine services had sufficient total hours of cases to warrant an OR allocation on at least 1 day of the week. The remaining two services were assigned to the OTHER service. The OTHER service allocation also contained cases from all services on days when the services were not allocated an OR (see Methods). In addition, one OR was allocated daily for urgent cases.

At room noise levels of 5% and 30%, the total amount of OR time allocated among all surgical services decreased to 0.4% and 4.8% of baseline, respectively (Table 1). The corresponding staffing costs increased by only 0.03% (upper 95% CI, 0.04%) and 1.4% (upper 95% CI, 1.7%), respectively.

Table 1

Table 1

The absolute reduction in allocated OR time was larger than the absolute increase in staffing costs. Also, over-utilized time was considered in the allocation process to be more expensive than regularly staffed and allocated OR time (1). Thus, most of the reduction in allocated OR time was OR time that was not being used in the baseline allocation. Otherwise, the absolute increase in costs would have exceeded the absolute reduction in allocated OR time.

The results were sensitive to the size of the surgical service. For services allocated 15 or fewer 8-h ORs per week, there were no changes in the number of allocated ORs for 0% up to a 30% level of room noise (Fig. 1). As the room noise level was increased from 0% to 30%, the largest-volume service (28 ORs per week) had a progressive decrease of 0 to 5 ORs allocated per week. The second largest volume service showed no decrease in allocated ORs until the room noise level reached 25%.

Figure 1

Figure 1

Back to Top | Article Outline

Discussion

The importance of our article is in showing that uncertainty in knowing the OR in which each case was actually performed has a very small effect on the total staffing cost of allocating ORs to surgical services, even at an uncertainty rate of 30%. At frequencies of room noise seen in practice (<10%), using these “noisy” data added <0.1% to the achieved staffing cost compared with using data in which the actual OR was known for every case.

In analyzing data from a number of facilities, we have observed that error rates in ORIS data tend to decrease once staff understand that important decisions are being made on the basis of these data. Therefore, whatever small increment in cost that there may be from the presence of room noise when facilities first begin to use their ORIS data to allocate ORs on the basis of maximizing OR time efficiency, this effect may decrease over time.

Our results suggest that most surgical suites will not need to research and correct errors in the actual OR recorded by the ORIS for purposes of OR allocation. Rather, when two cases are recorded as overlapping, the incorrectly recorded case can simply be reassigned to a unique “unknown” OR. We are not, however, suggesting that staff should be careless in recording the OR location or the OR entrance or exit times. There are other reasons (e.g., medicolegal considerations) why this information should be documented accurately. Furthermore, if the organizational culture does not encourage accurate entry for some data fields, that may engender carelessness for the entry of other, more critical, data. If there are many errors in OR recording, this may indicate that other important parts of the data are suspect too.

There are other common errors in ORIS data. Small errors in OR entrance and exit times are also common. Such errors in listed OR times will decrease or increase the total hours of cases and simultaneously increase or decrease the corresponding turnover times. The effect is a negligible change in the total hours of cases, including turnover times. OR allocations based on OR efficiency are calculated from each service’s total hours of cases, including turnover times. Therefore, OR allocations and the resulting staffing costs are resistant to small errors in OR entrance and exit times.

On the basis of our sensitivity analysis, we think that there are two reasons why uncertainty in knowing the actual OR in which each case was done had only a small effect on OR allocations and staffing costs. The first of the two reasons was that OR allocations were based on each surgical service’s total hours of cases, including turnover time. These total hours of cases, including turnover times, were not affected greatly by adjusting the OR, because the duration of surgery was longer than the turnover time. When a case that was actually performed as either the first or the last in a series of cases in an OR is reassigned to an unknown OR, the total hours of cases are decreased by one turnover time (because there is no turnover after the last case in an OR). For cases in an OR that have a preceding and a following case, two turnover times are lost. However, the turnover time between the remaining cases is increased between the two cases surrounding the reassigned case to the default maximum turnover time. In the dataset analyzed, the average turnover time was 37 minutes, with a default maximum turnover time of 60 minutes, whereas the average duration of surgery was 191 minutes.

There is a second reason why uncertainty in knowing the actual OR in which each case was done had only a small effect on OR allocations and staffing costs. ORs are allocated in whole multiples (e.g., zero, one, or two ORs per day). The effect of having to allocate 0 hours, 8 hours, 16 hours, and so forth, of OR time to a surgical service has a much larger effect than a small decrease in total hours of surgery, including turnover times. For example, because the cost of 1 over-utilized hour was considered to be equivalent to 1.75 under-utilized hours (1), a service’s allocation would be rounded to one 8-hour allocation if it averaged between 3.0 and 10.9 hours of surgery per day. Because the difference between 8 and 3.0 hours was larger than the difference between 10.9 and 8 hours, most of the OR time that was taken away from the baseline allocation was not being used. Therefore, the decrease in total hours of surgery was not matched by a corresponding increase in staffing cost.

Understanding the reason for the sensitivity of our results to the OR time allocated to surgical services is important for appreciating the broad implications of our findings. Only the two services receiving the largest number of daily OR allocations demonstrated any decrease in their allocations from introducing room noise into the data, and one of them only when the room noise level reached 25%. Because typical room noise levels are <10%, this suggests that for the majority of services at a facility, there will be little, if any, effect of such errors on OR allocations calculated with the algorithm that maximizes the efficiency of the use of OR time. Only for services with sufficient surgical volume to warrant an allocation of five or more eight-hour ORs per day do the turnover times lost add up to a sufficient number to result in a different allocation (Table 1).

The findings of our study are subject to several limitations. The first limitation was that the method that we used to introduce uncertainty in the ORs in which cases were done resulted in a random distribution of room noise among all surgical services. If, at a given facility, a particular service experiences a substantially more frequent rate of incorrectly recorded actual OR locations than other services, then the effect on the OR allocations to that service might be affected to a greater degree than predicted by this study. This situation could easily be detected by running a query against the ORIS database. Our analysis (Fig. 1) applies using the percentage of room noise for that individual service. The reason for this is that each service received its OR allocation independent of other services (1–3,9). Each service received the right amount of OR time for that service to maximize OR efficiency. Thus, 15% room noise added to all services achieved approximately 15% room noise to each service.

The second limitation was that the reduction in total hours of surgery from moving cases into an “unknown” OR is directly related to the average turnover time. At facilities with long turnovers, this may result in a higher cost of the “noisy” allocation, because the subsequent reduction in allocated ORs might cause more cases to be completed during over-utilized time. To mitigate this problem, a small amount of turnover time could be added to each case moved, to partially offset the decrease in total hours of cases.

The third limitation was that the staffing cost of a given level of room noise may be decreased at facilities that perform a large number of short cases in each OR. This is because a larger percentage of the cases that are moved will actually have been followed and preceded by another case. When this occurs, the decrease in turnover time from moving the case to the “unknown” OR will be offset somewhat by the increase in turnover time between those surrounding cases. When a case is the first or last case in the OR (assuming that more than one case was done in that OR), then the total hours of cases will be reduced by the turnover time that followed or preceded the moved case, respectively. However, in this situation, the conclusions of our study are strengthened, rather than weakened.

In conclusion, our study applies to surgical suites at which the surgeon and patient choose the day of surgery and OR time is allocated to each service based on maximizing the efficiency of use of OR time. We demonstrated that as much as 30% uncertainty in knowing the actual OR in which cases were performed has a minor effect on both staffing cost and OR allocations. In most circumstances, facilities with ORISs will be able to simply resolve this uncertainty by considering such cases as if they were done as the only case in a different, “unknown” OR.

Back to Top | Article Outline

References

1. Dexter F, Epstein RH, Marsh HM. Statistical analysis of weekday operating room anesthesia group staffing at nine independently managed surgical suites. Anesth Analg 2001; 92: 1493–8.
2. 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.
3. Strum DP, Vargas LG, May JH. Surgical subspecialty block utilization and capacity planning: a minimal cost analysis model. Anesthesiology 1999; 90: 1176–85.
4. Dexter F, Traub RD. Determining staffing requirements for a second shift of anesthetists by graphical analysis of data from operating room information systems. AANA J 2000; 68: 31–6.
5. Dexter F, Traub R. 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.
6. Dexter F. A strategy to decide whether to move the last case of the day in an operating room to another empty operating room to decrease overtime labor costs. Anesth Analg 2000; 91: 925–8.
7. Dexter F, Macario A, O’Neill L. A strategy for deciding operating room assignments for second-shift anesthetists. Anesth Analg 1999; 89: 920–4.
8. Epstein RH, Dexter F. Statistical power analysis to estimate how many months of data are required to identify operating room staffing solutions to reduce labor costs and increase productivity. Anesth Analg 2002; 94: 640–3.
9. Dexter F, Macario A. Changing allocations of operating room time from a system based on historical utilization to one where the aim is to schedule as many surgical cases as possible. Anesth Analg 2002; 94: 1272–9.
Back to Top | Article Outline

Attention Authors!

Submit Your Papers Online

You can now have your paper processed and reviewed faster by sending it to us through our new, web-based Rapid Review System. Submitting your manuscript online will mean that the time and expense of sending papers through the mail can be eliminated. Moreover, because our reviewers will also be working online, the entire review process will be significantly faster. You can submit manuscripts electronically via http://www.rapidreview.com. There are links to this site from the Anesthesia & Analgesia website (http://www.anesthesia-analgesia.org), and the IARS website (http://www.iars.org). To find out more about Rapid Review, go to http://www.rapidreview.com and click on “About Rapid Review.”

© 2002 International Anesthesia Research Society