Time management in operating rooms ( ORs ) – a retrospective study to evaluate estimated and objective durations of surgical procedures in the �

Lena Zaubitzer Universitatsklinikum Mannheim Annette Affolter Universitatsklinikum Mannheim Sylvia Büttner Universitatsklinikum Mannheim Sonja Ludwig Universitatsklinikum Mannheim Nicole Rotter Universitätsklinikum Mannheim Claudia Scherl Universitätsklinikum Mannheim Sonia von Wihl Universitatsklinikum Mannheim Weiß Christel Universitatsklinikum Mannheim Anne Lammert (  anne.lammert@umm.de ) Universitatsklinikum Mannheim


Background
Between 1995 and 2006, health expenditures increased 2.5% per year. The most cost-intensive institutions are hospitals. Personnel and material costs of hospitals particularly increased between 1995 and 2006, which both rose by approximately 39% [1]. Health expenditures are increasingly nanced by private households and private organizations. Furthermore, investment in hospitals by federal states decreased by over 20% between 2001 and 2011 [1]. These circumstances can only be counteracted by cost minimization, increasing revenues and process optimization. The operating department is an essential part of surgical patient care, but, along with intensive care units, it is also a very cost-intensive division. However, it is one of the main sources of revenues of hospitals [2,3]. The objective of operating room (OR) management must be optimal use of this important source of revenue. The three most important resources in OR management are the OR itself, time and personnel. Their utilization must be optimized to realize maximal pro t by operating departments [4]. Thus, one of the most important parts of OR management is the planning of surgery durations. An e cient utilization of ORs can only be assured by accurate planning of OR times. Ine ciency is caused by diverging surgery durations (either too long or too short), which result in rescheduling, idle time and overtime [5], leading to increased personal costs and patient dissatisfaction. This leads to overtime hours for the surgeons as well as other occupational groups and speci c elds that have to work overtime in consequence of inaccurate planning. This in turn has the potential for con icts between different team members in the OR. Studies also revealed that different persons on a surgical team partly represent various interests and pursue different goals. This is also caused by the fact that a surgical team is composed of different individuals with divergent values. Moreover, it is a multidisciplinary group of people with different organizational structures within individual occupational groups. Furthermore, e ciency and economy are not among the most important goals of every team member. All these factors abet con icts. Con icts in turn result in employee dissatisfaction and hence reduced productivity because employee satisfaction correlates with productivity [6]. Therefore, good planning is important to reduce potential con icts. OR planning is made on the basis of estimated durations. It must be noted that the OR scheduler is in uenced by many factors and mostly does not act on economic rationality [5]. The planner sometimes does not know the patient personally. In consequence, important facts potentially in uencing operating times may be missing. It is important that the planner obtains information about the estimated operating time from the surgeon who better knows the patient. Moreover, non-technical skills that are important for surgeons, such as communication, preparation and planning, are sometimes missing. All these skills are important for a successful surgery as well as for accurate planning [5,7].
There are many potential sources of errors in surgery scheduling. The main objective of our research was to disclose general false estimation of durations. There are different time durations in ORs, which have to be taken into account. First, there is the duration between the patient entering the OR and the surgeon's arrival. Then, there is the setup duration, which is the time between the patient`s arrival at the OR and the start of surgery. Included in the setup time is surgical preparation, i.e., the time from the beginning of the patient's preparation, which is performed by the surgeon, to the start of surgery. The incision to suture time de nes the duration between the incision (for a surgical procedure) and the end of the suture. After this, the post-processing time begins. It ends when the patient leaves the OR.
Here, we also included surgical post-processing time, which is the time from the end of the suture to the end of all surgery-related activities done by the surgeon. Another important time encompasses the duration between the patient entering and leaving the OR [8].
Moreover, the practical experience of the surgeon has to be considered. Recent studies have already revealed that surgeries are signi cantly longer with residents being involved [9]. It is generally accepted that surgical training is associated with increased surgical duration. Training the next generation of surgeons is a fundamental part of an effective health care system. Teaching surgical residents is associated with decreased operative e ciency owing to the time needed to give instructions by a specialist and due to slower operative speeds of trainees. The duration of a surgical residency, which takes several years, can result in substantial cumulative ine ciency, impacting the cost as well as access to limited surgical resources. Thus, another aim of our study was to determine which kind of operating times are prolonged when operations are conducted by residents. For the supervising specialists, it might be interesting to know for which surgical procedures residents need more time. This information can help save time by better supporting residents during these procedures.

Methods
In a retrospective study, time durations of 1809 operations under general anaesthesia were analyzed. All surgeries were carried out between the 1st of January and 31st of December 2018 at the ENT Department Moreover, estimated operating times were requested of 10 specialists and 17 residents with the aid of a questionnaire. Three specialists did not receive a questionnaire or did not answer it for any reason. One specialist who lled out the questionnaire did not perform surgery in the daily routine but was involved in operating time scheduling. Two residents could not obtain a questionnaire, because they had left our department in the meantime. Surgeons estimated their own operating room times, average operating room times of all surgeons, and set-up and post-processing times for the different types of surgery mentioned above. Then, times estimated by specialists were compared to the residents' estimations. Moreover, the median values of estimated durations were checked against the median value of effectively taken (objective) operating room times. These comparisons have been done separately within the group of Rs and within the group of Ss.
For all comparisons between two groups, Mann Whitney U test has been used. A p-value less than or equal to 0.05 was considered statistically signi cant. All statistical analyses were conducted with the aid of SAS software, release 9.4 (SAS Institute Inc., Cary, North Carolina, United States).
The study was approved by the ethics committee II of Ruprecht-Karls-University, Heidelberg, Medical Faculty Mannheim.

Results
The comparison of objective OR times of residents and specialists showed signi cant differences for various types of surgeries (tables 1 and 2). There was a signi cant difference in surgical post processing time after the extirpation of a cervical lymph node between residents and specialists (p = 0.006), with residents taking signi cantly more time. Time between the patient s entrance into the operating room was somewhat longer for the residents (p=0.030).
Comparing the OR times of panendoscopy including microlaryngoscopy, signi cant time discrepancies between residents and specialists could be ascertained for the time between the surgeon's entrance and the patient's entrance into the operating room (p= 0.007). Differences were also observed in surgical preparation time (p<0.001), incision to suture time (p<0.0001), post-processing time (p=0.025), entire operating room time (p=0.001), entire duration of surgical action (p= 0.012) and the sum of incision to suture time and time of surgical action (p<0.0001).
Except for arrival time to the OR and post-processing time, durations of specialists were signi cantly shorter than those of residents.
Regarding sinus surgery, the study revealed signi cant time differences regarding the time between entering of the surgeon and patient in the OR (p=0.018), incision to suture time (p<0.0001), entire OR time (p<0.0001) and the sum of incision to suture and surgical preparation and post-processing time (p<0.0001). As before, residents were more on schedule when they entered the operating room, but they operated slower than the specialists.  and time of surgical action (p<0.0001). All time durations were signi cantly longer when the surgery was performed by residents.
In addition, the time between entering of the surgeon and patient into the operating room (entrance time) (p=0.009), setup time (p=0.007), incision to suture time (p=0.002), post-processing time (p=0.015), whole operating room time (p=0.004), time of surgical action (p=0.040) and the sum of incision to suture time and time of surgical action (p=0.001) differed signi cantly for tonsillectomy with uvulopalatopharyngoplasty depending on whether it was performed by residents or specialists. In addition, residents entered the operating room earlier but took longer surgery times than specialists.
Analyzing operating room times of intracapsular tonsillectomies, residents took signi cantly more time for incision to suture (p<0.0001), whole OR time (p<0.0001), and the sum of incision to suture and surgical action (p<0.0001).
Finally, there was a signi cant difference in the post processing time after lateral parotidectomies of residents compared to specialists (p=0.002). In this case, post-processing times of specialists were longer.
For some surgical procedures (Midface revision after bone fractures, submandibulectomy and tracheostomy) no statistically signi cant differences could be found. This may partly be attributed to the rather small sample sizes of the relevant subgroups.
In summary, specialists often take more time until they enter the OR, but then they perform surgeries faster than residents.
Residents and specialists also differ sometimes in their time estimates (table 3). When asked about the average incision to suture time of all surgeons, estimations were signi cantly different in the case of implantation of hypoglossal nerve stimulator (p=0.034), midface revision after bone fractures (p=0.010), microlaryngoscopy with panendoscopy (p=0.028), tonsillectomy and lateral parotidectomy (p=0.036).
Estimated durations of preparation time by residents and specialists only differed signi cantly in terms of implantation of a stimulator of hypoglossus nerve (p=0.021) and total parotidectomy (p=0.040).
Specialists always estimated longer durations than residents, except for their own OR time. Concerning their own OR time, specialists expected less time than residents for septoplasty, tonsillectomy, tracheotomy and tonsillectomy with uvulopalatopharyngoplasty.
A comparison of the estimated and effectively measured OR times by specialists revealed biased estimations, especially of preparation time and incision to suture time, and rarely also of post-processing time. Specialists underestimated post-processing time only for panendoscopy including microlaryngoscopy (p=0.015), and for septorhinoplasty (p=0.007). Incision to suture time of most surgeries was overestimated by specialists. The median expected incision to suture time was signi cantly higher than that measured in case of extirpation of a cervical lymph node (p=0.024), implantation of hypoglossal nerve stimulator (p=0.039), midface reconstruction after bone fractures (p=0.008), panendoscopy including microlaryngoscopy (p=0.001), sialendoscopy (p=0.046) and tonsillectomy (p=0.001 In summary, specialists poorly estimated post-processing time in two out of 22 types of surgeries, incision to suture time in seven out of 22 types and preparation time in 16 out of 22 types of surgeries.
In the case of signi cantly false estimations, residents generally underestimated overall time.
In summary, residents underestimated objective preparation and post-processing time in seven out of 15 types of surgeries and incision to suture time in three out of 15 different types.

Discussion
The study revealed that the experience of the individual surgeon who will perform the surgery must be considered when scheduling surgical procedures. Thus, specialists sometimes enter ORs with delay but then conduct the surgery faster than residents. Possible reasons for the specialists` delayed entrance into the OR are that specialists often have multiple responsibilities such as managing a hospital ward parallel to their consultation hours. Prior examinations already showed that absence of the surgeon is a frequent reason for delays in start time [10]. Therefore, for OR planning, other duties of the intended surgeon, such as meetings and consultation hours, must be considered. Moreover, communication, in terms of feedback in case of being late and reminders of being on time, seems to be a strategy to prevent delay of surgeons [11,12]. As a consequence of less routine and practical experience, residents take more time for surgeries. Thus, the surgeon's operating experience should always be respected in OR scheduling. Former studies have ascertained that the presence of residents in the OR results in longer durations of surgeries and increasing costs. The training of residents takes time, however being aware of this fact might help to improve time estimates of planned surgeries [9]. One possible method to save time in ORs may be the improvement of surgical education outside of the OR. Academic hospitals should invest in simulator training programs for residents. Former studies already revealed, that simulator surgery trainings improve the performance of residents in the OR. After simulator training residents operate faster than the control group without training [20][21][22].
The differences in estimated OR times between residents and specialists can also be explained by different experiences of these subgroups. Residents seem to be more optimistic concerning estimated average incision to suture time, post-processing time and preparation time. Only regarding their own individual duration of surgeries, did specialists expect shorter time periods than residents. Specialists calculate less operating time in six out of 15 types of surgeries. In 4 out of these 6 types of surgeries, their objective OR times are in fact shorter than those of residents. Those four types (nasal septoplasty, tonsillectomy, tonsillectomy with uvulopalatopharyngoplasty, intracapsular tonsillectomy) are typically used for teaching and are often performed by residents. Therefore, it can be hypothesized that specialists are aware of the fact that residents take more time. However, this knowledge does not guarantee that this fact is always taken into account in OR scheduling. There may be a behavioral bias of the planning surgeon.
Predicting the duration of a surgical procedure is somewhat similar to predicting the duration of a sports game. Although you might know that a soccer game takes 90 minutes, it is not possible to know exactly how long the game will last. Furthermore, surgical procedures can go into overtime if unexpected complications or ndings occur causing adaptations of the surgical procedure, which requires extra time.
On the other hand, there are several factors reducing the expected duration of a surgical case. As the precise estimation of operating times is a prerequisite for e cient OR scheduling, it is important to determine if OR times estimated by surgeons are comparable to "real life" OR times. Our ndings suggest that assessments by residents and specialists are reasonably acceptable in terms of cut-seam times, but they often underrate the necessary resources (time, personal) when evaluating times of preparations before cutting and times of post-processing. In our study, preparation times before incision were objectively twice, almost three times, as long as estimated in many types of procedures, e.g., cochlear implant and implantation of hypoglossal nerve stimulator, but so were "daily routine" procedures, such as septorhinoplasty or endonasal sinus surgery (Table 1). However, how should they know? They enter the OR for cutting and typically leave after cutting. When surgeons think of a surgical procedure, they mainly think of the process they are taking part in and lose sight of the needs before and after the "surgical act." Wright et al. sought to compare surgeons' time estimates for elective cases with those of commercial scheduling software to assess whether improvements could be made by regression modeling. Encouragingly, surgeons in general topped the commercial scheduling software in scheduling OR times. Individual surgeons were even better compared to the software. A simple model combining surgeons' estimates with the historical data signi cantly reduced the prediction errors. They also determined that surgeons need an incentive to reduce their errors in estimating duration. As the impact of managed care grows, the incentive to be more accurate may become greater [13]. Whatever system is used to schedule OR times, having more precise estimates of each case's duration and emergence of resources should help reduce underutilization and excess planning of the workday [13][14][15].
With our data we demonstrate for the rst time, that some surgical procedures in ENT are easier to estimate than others, e.g., preparation times, cut-seam times and post-processing times of a tonsillectomy or a septoplasty are easier to estimate than durations of a complicated oncological resection of the head and neck. This corresponds with the ndings of Gordon et al., who noted that laparoscopies (the second-most-frequent procedure) at John Hopkin`s Medical Institutions varied by 42% from the estimated time, whereas hysterectomies, inguinal hernia repairs, and prostatectomies varied by up to 4% [16]. Our data support the contention that some types of surgeries are inherently di cult to predict.
However, we analyzed nearly 2000 procedures, which allowed us to sustain the mean duration of speci c procedures. Despite unpredictable events, such as management problems, patients' transport to the OR or complications/di culties that may occur during an operation, management and scheduling of the OR must be done based on that kind of objective information [17,18]. The duration of a surgical case depends on patients' individual characteristics of the disease and the surgeon's skill and routine with the procedure performed. This can be shown by the differences in time needed for an ENT resident compared to a specialist (Fig. 1). Vinden et al. determined that a wide range of surgical procedures require signi cantly more time to perform in teaching than in nonteaching hospitals. They note that the magnitude of this difference is large enough to potentially affect direct and indirect costs, institution and surgeon e ciency, and possibly impact surgical outcomes [19].

Conclusions
Our data suggest that surgeons do a good job in estimating and scheduling surgical procedures. However, surgeons tend to lose sight of resources and the time necessary to prepare for and perform follow-up course work after surgical procedures. In daily work, routine OR scheduling is usually based on surgeons' appreciations of cut-seam times without realistic perspectives on resources needed for activities before and after the surgery. To avoid errors of estimation regarding pre-and postoperative durations, anesthesiologists and OR-assistants (nurses) should be more involved in OR scheduling to avoid excess planning or waste of OR capacity. Competing interest: There are no competing interests ( nancial/non-nancial) of the authors.
Funding: There was no funding regarding this study.
Authors` contributions: LZ was a major contributer in writing the manuscript AA was a major contributer in reading/correcting and improving the manuscript SB was a major contributer in analyzing data statistically and contributed Fig. 1 SL was a major contributer in reading/correcting and improving the manuscript NR was a major contributer in reading/correcting and improving the manuscript CS was a major contributer in reading/correcting and improving the manuscript SvW analyzed and interpreted the scheduling data CW was a major contributer in analyzing data statistically and contributed Tables 1-3 AL was a major contributer in writing the manuscript