| JOURNAL HOME | CME HOME | THIS MONTH | PAST ISSUES | ETOC | COLLECTIONS |
| AUTHORS | REVIEWERS | EDITORIAL BOARD | FEEDBACK | RSS | HELP |
| ||||||||||||||
| ||||||||||||||
|
|
|
|
||||||||||||
|
||||||||||||||
Departments of Anesthesia,
*University of Iowa, Iowa City, Iowa, and
Stanford University, Stanford, California;
College of Business Administration, North Dakota State University, Fargo, North Dakota
Address correspondence to Franklin Dexter, MD, PhD, Department of Anesthesia, University of Iowa, Iowa City, IA 52242. Address e-mail to franklin-dexter{at}uiowa.edu
 |
Introduction |
|---|
 Top Introduction 1. Logic Behind Using...2. Pitfalls that May...3. How Using Information...4. Conditions Under Which...5. Review of Management...6. Recommendations in...References |
|---|
The next generation of process improvement efforts may come through deployment of information technology. For example, information systems are being developed and implemented to coordinate the scheduling of patient appointments among different sites within a health care system. The challenge in using these information systems is to decrease overall costs by scheduling patients to match optimally available providers, while simultaneously improving customer service, including patient access to appointments and ease of making those appointments.
Enterprise-wide patient scheduling systems (EWPS) permit clerks, nurses, physicians, or patients (e.g., via the Internet) to schedule patient appointments throughout a multiple-specialty physician group and/or health care system. An application of these EWPS in perioperative medicine is the coordination of patient scheduling in surgeons’ clinics with patient scheduling in surgical suites. Better coordination of clinic and operating room (OR) schedules is an important objective for health care systems. This is because most costs related to delivering surgical services are for staff (1). In addition, patient satisfaction with health care systems relies in part on the ease of obtaining timely appointments (2).
Depending on the structure of the health care system, the OR staff costs may include not only wages for OR nurses, but also for surgeons and anesthesiologists (e.g., at state-funded hospitals) (1). Perioperative costs can be reduced if cases are scheduled such that the workload evenly matches scheduled staffing. Specifically, to minimize the cost per case, "downtime" must be minimized. Because evenly matching workload and staffing is important, properly choosing the day on which each patient will have surgery (3) is the most important decision affecting OR labor costs.
A second factor in evenly matching OR workload and staffing is how long patients have to wait for their surgeries. Patients generally want to have surgery as soon as possible (4). Thus, by better managing surgical patient scheduling, health care systems can decrease the cost per case (i.e., increase OR utilization) while maintaining or even improving patient satisfaction.
The goal of this article is to review the management sciences literature to assist health care systems in programming EWPS information systems so that they can be used to increase OR efficiency. The article is divided into six sections that review:
|
1. Logic Behind Using EWPS to Benefit Both Patients and the Health Care System |
|---|
|
TopIntroduction1. Logic Behind Using...2. Pitfalls that May...3. How Using Information...4. Conditions Under Which...5. Review of Management...6. Recommendations in...References |
|---|
Example 1
A pulmonologist sees a patient and diagnoses biopsy-proven lung cancer. The pulmonologist recommends to the patient that the patient consult a surgeon to undergo lobectomy (Figure 1). The patient asks the pulmonologist to recommend a surgeon. The health care system in which the pulmonologist works has two general thoracic surgeons: Dr. Jones and Dr. Smith. The pulmonologist routinely refers patients to both surgeons.
|
In contrast, Dr. Smith cannot see the patient in clinic until two weeks from now. However, Dr. Smith has open OR block time to schedule the case for the next several weeks.
The scheduling question is whether the patient should be referred to be cared for by Dr. Jones (who has an earlier clinic opening) or Dr. Smith (who has earlier open OR time). EWPS software would provide this information to the pulmonologist and to the clerk responsible for scheduling the patient. Although Dr. Smith has the longest clinic wait, referring the patient to Dr. Smith would minimize the length of time the patient has to wait to have surgery and improve utilization of the surgeons’ OR block time. This example illustrates the potential attractiveness of EWPS software.
Note that in the above example, the EWPS software does not actually make the scheduling decision. Instead, EWPS provides the schedules for Drs. Jones and Smith so that patients and staff can make a scheduling decision. The assumption in providing the schedules is that by having access to the appropriate data, patients and staff can make better scheduling decisions. The conditions under which this is actually true are unknown.
|
2. Pitfalls that May Occur with the Implementation of EWPS to Integrate Surgical Clinic and OR Scheduling |
|---|
|
TopIntroduction1. Logic Behind Using...2. Pitfalls that May...3. How Using Information...4. Conditions Under Which...5. Review of Management...6. Recommendations in...References |
|---|
Example 2
Two surgeons are partners belonging to the same multiple-specialty group. Each surgeon operates on one patient each and every day of the five-day workweek. The two surgeons also see patients in clinic each workday. Four scheduling rules for these two surgeons were used to create the corresponding Figures 2 and 3:
|
|
Under the above rules, if patients are able to get clinic appointments on the same days that they are referred to the surgeons, then after several weeks, all patients will be waiting 10 workdays to have surgery (10 days = [5 patients + 15 patients] ÷ [2 surgeons x 1 case per day]) (Figure 2). This means that the delay between the time when the patient is referred to the surgeons in the multiple specialty group until the patient is being seen by a surgeon is zero days.
Next, the rules used to create Example 2 and Figure 2 were changed to create a new, more realistic, scenario. Surgeons’ clinic schedules are full with so many patients that the time required to find an open clinic appointment is increased from 0 days to 10 days. Under this condition, the simple rules of Example 2 produce large oscillations in the numbers of patients waiting in each surgeon’s OR queue and in the lengths of time patients wait for surgery (Figure 3).
Thus, if patients are able to get a clinic appointment immediately, the numbers of patients in the OR queues even out (Figure 2). If patients have to wait for a clinic appointment, a sustained oscillation occurs in the numbers of patients waiting in each surgeon’s OR queue and in the lengths of time patients wait for surgery (Figure 3). Both patient satisfaction and OR efficiency will suffer, as explained below.
In Example 2, each surgeon performs one case per day. If the scheduling clerk, patient, or referring physician uses data from the EWPS system about the surgical queue to schedule patients into surgical clinics, the length of time that patients wait for surgery can oscillate (Figure 3) (5).
In Figure 3, the length of time (i.e., number of days) that patients wait for surgery oscillates. Equivalently, if the surgeon adjusts the number of cases that he or she performs each day to maintain some reasonable length of time that patients wait for surgery (4), then the OR workload would oscillate. This variation in OR workload directly influences the efficiency of the OR. When the workload is increased, the OR manager schedules additional OR hours to handle this (short-term) increase. Then, when the OR workload diminishes, the staffing level is too large, and OR utilization decreases.
Alternatively, the OR manager may recognize the statistical challenges (4) in adjusting OR staffing levels to match short-term changes in the length of time that patients are waiting. The result is that patients’ waiting times will increase, likely reducing patient satisfaction. These patients will generally view the OR as being less efficient. In addition, high OR utilization may cause the surgeons to complete more operations outside of block time, thereby increasing staff costs (3,6). Thus, variation in OR workload affects the efficiency of the OR in different ways.
In the next section, it is clarified why a delay between the time when the patient is referred to the surgical group until the patient is seen by a surgeon can cause oscillations in three end points: number of patients in the OR queues, average length of time that patients wait for surgery, and number of OR cases per day. In this article, these three, together, are referred to as OR workload.
|
3. How Using Information About Availability of Surgeons’ Open OR Time to Schedule Clinic Appointments as Part of EWPS Can Impair OR Efficiency |
|---|
|
TopIntroduction1. Logic Behind Using...2. Pitfalls that May...3. How Using Information...4. Conditions Under Which...5. Review of Management...6. Recommendations in...References |
|---|
Example 3
A clerk works for a multiple specialty group practice in which patients are usually referred to the health care system and only rarely are referred to individual surgeons. Let us assume that each day the clerk schedules many patients into several surgeons’ clinics. One of the surgeons is Dr. Chang. When Dr. Chang’s OR schedule is "busy," defined as more than 12 patients booked for surgery, the clerk schedules new patients into other surgeons’ clinics. When Dr. Chang’s OR schedule is "open," defined as fewer than 12 patients booked for surgery, the clerk schedules patients into Dr. Chang’s clinic.
Let us further assume that Dr. Chang averages one surgical case per day. However, on Day 3, Dr. Chang happens to perform two cases, not just one (Figures 4 and 5). On Day 4, the clerk identifies that the number of patients in Dr. Chang’s OR queue has decreased by two. The clerk schedules two patients to be seen by Dr. Chang. The patients are seen in Dr. Chang’s clinic on the day that they are scheduled into Dr. Chang’s clinic (i.e., patients enter Dr. Chang’s OR queue with no delay). By Day 5, the number of patients in the OR queue has increased back to 12. The number of patients in the OR queue from then on remains a constant. This first scenario of Example 3 does not include any delay that patients may experience in getting a clinic appointment, meaning that they would not be seen by a surgeon on the day that they are referred to surgeons in the multiple specialty group.
|
|
|
The conclusion drawn from the foregoing analysis is that because of the scheduling policy, one additional patient having surgery on one day can result in a sustained oscillation in OR workload (i.e., the number of patients referred to a surgeon, the length of time that patients wait for surgery, and utilization of the surgeon’s OR time) (Figures 3 and 6). This occurs because the time delay between when clinic appointments are assigned and the actual appointments leads to the excessive referral of patients to the surgeon who supposedly has available OR time.
|
4. Conditions Under Which EWPS Will Paradoxically Cause Adverse Oscillations in OR Workload |
|---|
|
TopIntroduction1. Logic Behind Using...2. Pitfalls that May...3. How Using Information...4. Conditions Under Which...5. Review of Management...6. Recommendations in...References |
|---|
If a scheduling clerk uses data on surgeons’ clinics and OR queues to recommend a clinic date and surgeon to a patient, the clerk is ordering inventory (patients). When the clerk reviews the number of patients in a surgeon’s OR queue, the clerk is functioning like a clerk in a store who is reviewing the amount of supplies in inventory. The clerk schedules patients into a surgeon’s clinic similarly to a store clerk ordering supplies. The delay from when a patient is scheduled into a surgeon’s clinic until the patient is booked for surgery is analogous to the time it takes for a distributor to deliver supplies to the store. Just as the delay impacts inventory management, the delay affects the size of OR queues (7).
The oscillations occur in Figures 3 and 6 because the clerk orders new inventory (patients) each day based on the current inventory (patients waiting for surgery) while disregarding the items that have been previously ordered but have not yet arrived at the store (patients waiting to be seen in the surgeons’ clinics). The oscillation in the OR queue is larger in Figure 3 than in Figure 6, because in Figure 3 there is a longer delay between when patients are referred to surgeons until the patients are in the surgeons’ OR queues and so are observed by the clerks.
In the above examples, an EWPS provides clerks (or patients, nurses, physicians, etc.) with information about surgeons’ clinics and OR queues. Clerks use this information to schedule patients into surgeons’ clinics. If the clerks disregard these new patients when the size of the surgeons’ OR queues are considered, oscillations will occur (i.e., there is no uncertainty—they will occur). The magnitude of the oscillations will vary depending on the duration of the delay encountered.
The tendency for those managing inventories to disregard inventory previously ordered and keep on ordering more is a common teaching point in management science (7). The longer the delay, the greater the chance of a clerk "forgetting" about a patient (previously ordered inventory) that he or she has previously scheduled. This results in oscillations in ordering (patients scheduled into a surgeon’s clinic) and inventory (number of patients in the surgeon’s OR queue) (7). Studies show that people working manually with data almost always underestimate the delay between placing and receiving orders, resulting in over-ordering and oscillations in supply and inventory (7). Mathematical analysis and computer support are needed to prevent this problem.
Oscillations in OR workload are likely to occur in perioperative care settings. This is because it may be challenging for scheduling clerks to remember why each patient was assigned to be evaluated by one surgeon or another. Clerks in large health care systems may schedule hundreds of patients into dozens of physicians’ clinics. As such, these clerks may not remember (or may not know) that a patient has been scheduled weeks beforehand into a surgeon’s clinic to increase a surgeon’s OR queue.
A second reason that the oscillations may be particularly important in hospitals with EWPS is that many patients scheduled into surgeons’ clinics do not have surgery and thus leave the system having never been in an OR queue. This is in contrast to traditional inventory management in which supplies do not generally disappear. Keeping track of an inventory of patients may be more difficult than an inventory of supplies.
Third, oscillations may affect health care systems because there are multiple scheduling clerks. Unless clerks communicate among themselves that a patient has been scheduled into a surgeon’s clinic to make up for a current shortfall of patients in the surgeon’s OR queue, oscillations will occur. Yet, one purpose of having an EWPS information system is so that a patient does not have to speak to multiple individuals to be scheduled into a surgeon’s clinic. Typically, the clerk in the internist’s clinic would use data provided by the EWPS system to schedule the patient into an appropriate surgeon’s clinic.
A fourth reason to expect oscillations in OR workload is if patients make their own appointments (e.g., via the Internet). This article describes a traditional scenario in which clerks use EWPS information systems to schedule patients. However, costs may be lower and patients may prefer to use the newly released EWPS products (e.g., Medicware, Duarte, CA) in which patients use the Internet to schedule their own appointments. Because patients will not know that other patients have been scheduled beforehand into a surgeon’s clinic based on the surgeon’s OR queue, oscillations are expected.
For the reasons described above, if there is more than one surgeon with open clinic time to which a patient can be referred to, and OR queue data are available to the clerks (or to patients booking their own appointments), oscillations in OR workload will result. In that OR staffing is generally a fixed cost on a day to day basis (3,6), such oscillations will decrease OR efficiency and increase perioperative costs.
|
5. Review of Management Sciences Literature Relevant to Finding a Solution to the Problem of Oscillations in OR Workload Caused by EWPS |
|---|
|
TopIntroduction1. Logic Behind Using...2. Pitfalls that May...3. How Using Information...4. Conditions Under Which...5. Review of Management...6. Recommendations in...References |
|---|
The examples given in Figures 2–6 understate the mathematical complexity in giving scheduling clerks data both on surgeons’ clinic and OR queues. The examples and figures in this paper do not factor in random (stochastic) behaviors, which affect patient scheduling (3). These random variables include the probability distributions of the times between patients’ requests to be scheduled into several different surgeons’ clinics, percentages of patients who are scheduled into surgeons’ clinics that do not have surgery, and probability distributions of surgical case durations. Appropriately programmed EWPS software would have to calculate the expected values of the surgeons’ OR queues in the future based on current queue lengths and these random variables. The software would also have to include mathematical consideration of the surgical clinic scheduling rules and the alignment of clinics and OR days. This review of the state of the art in research on stochastic control of networks of queues found that currently available techniques are inadequate to address these issues (8–11).
Engineering (mathematical) techniques have been developed for controlling management systems similar to that of Example 3 (Figure 6). Yet, these optimal policies were published just three years ago and rely on assumptions which markedly oversimplify the problem of coordinating surgical clinic and OR scheduling (8). These assumptions include that the probability that a patient will request to be scheduled in a given surgeon’s clinic at any given time is the same throughout the day and among days of the week. A second assumption is that the probability that the surgeon will be performing surgery on a patient is the same throughout the day and among days of the week, and that surgical cases are not performed one right after the other. Third, the delay from the referral of a patient to a surgeon until the patient decides to be scheduled for surgery is assumed the same for all patients. Because these assumptions do not reflect the realities of patient scheduling, the practicality of available state-of-the-art techniques for controlling perioperative scheduling systems is questionable.
The problem of the allocation of patients to one of two parallel queues was considered (e.g., open OR time for the each of the two surgeons in Example 2, Figures 2 and 3). This scheduling problem has been evaluated mathematically by others, albeit subject to simplifying assumptions. One assumption is that the probability that at any given time a patient will be requesting to be scheduled into a surgical clinic is a constant. Also, it is assumed that the probability that at any given time a patient will be having surgery is a constant. Subject to these assumptions, and provided that each patient is seen by a surgeon in clinic as soon as the patient’s clinic appointment is scheduled, the optimal algorithm to minimize the mean patient waiting time and maximize OR utilization is to refer each patient to the surgeon who has the fewest number of patients waiting for surgery (9).
If the delay between the referral of a patient to a surgeon until the patient is seen in clinic by the surgeon is increased to one work day, then the optimal policy to minimize mean patient waiting time and maximize OR utilization is to refer each patient to the surgeon with the fewest expected number of patients in the surgeon’s OR queue one day in the future (9). If the number of days the patient waits for a clinic appointment is longer than one work day, then scheduling the patient into the clinic of the surgeon with the shortest expected OR queue is not optimal (9).
This result means that, even subject to markedly simplifying assumptions, scheduling each patient (who is definitely going to have surgery) into the clinic of the surgeon who can perform surgery the soonest will not minimize the average patient wait for surgery and will not maximize OR efficiency.
Strategy if Each Surgeon Has a Panel of Patients to Follow
A different class of algorithms to control networks of queues was also considered in this paper. These algorithms focus on the entrance and exit of patients from the queues (Figure 7). These "container systems" (10) can be used to control patient scheduling in surgeons’ clinics and OR queues, subject to the assumptions that all patients referred to a surgeon choose to have surgery (although not necessarily after just one clinic visit) and that all such surgical cases are of the same duration (see below).
|
A container system can be used to coordinate surgical clinic and OR scheduling. In Figure 6, Dr. Chang saw patients in her clinic the day after each patient was scheduled into the clinic. This delay of one day between when the clerk scheduled the patient to see Dr. Chang and the patient entered Dr. Chang’s OR queue caused oscillations. By using containers, on Day 3, two virtual "containers" would have become empty, corresponding to two patients having surgery (Figure 7). On Day 4, the clerk would then schedule two patients (linked to two containers) into Dr. Chang’s clinic. With the one-day delay, the two patients would enter the queue on Day 5.
The important concept is that by using the container system, the disturbance on Day 3 would not cause an oscillation in queue size in Figure 7 as it did in Figure 6, despite the same delay of one day. The reason that the container system prevents oscillations is that with the container system the focus is on maintaining a constant number of patients in the surgeon’s panel. The container system does not just focus on the OR queue when assigning clinic appointments.
Using the container system approach to address unwanted oscillations in workload has three advantages over using the mathematical theory, based on probability distributions of queues, described in the preceding section. First, with a container system the probability distribution for the time between patient referrals to a surgical group does not need to be specified. Second, the delays corresponding to surgeons’ clinics’ queues can vary among surgeons and over time. Third, coordination of the days of the week between clinic and OR days does not need to be included in the analysis.
Unfortunately, the container system requires that two fundamental but artificial assumptions be made.
The container system can work well if all patients seen in the surgeons’ clinics either proceed to surgery or stop being cared for by the surgeon within a short period of time. If a surgeon follows patients for several clinic visits over a long period of time relative to the average waiting time for surgery, the container system may not work. This is because a container would be "filled" and held by the patient who is not having surgery, resulting in a decrease in the number of patients in the surgeon’s OR queue. Because many surgeons see patients who are not about to have surgery, this limitation to the container system is important and means that the container system is unlikely to be useful to EWPS.
The container system also will not work if a surgeon’s case durations differ substantially among patients, which is common. If case durations differ among patients, a container is no longer a measure of future work. Alternatively, the containers could correspond to the number of hours of OR time available to the surgeon, instead of to the number of patients cared for by the surgeon. This system also will have problems, because it is not generally known, at the time when a patient is scheduled into a surgeon’s clinic, how long the patient’s surgery will last. In other words, it is not known how many containers to assign to the patient. For this reason too, the container system is not suitable for preventing EWPS from causing impairment of OR efficiency.
|
6. Recommendations in Programming EWPS Software to Avoid Reductions in OR Efficiency |
|---|
|
TopIntroduction1. Logic Behind Using...2. Pitfalls that May...3. How Using Information...4. Conditions Under Which...5. Review of Management...6. Recommendations in...References |
|---|
EWPS is novel for many health care systems because Intranet-based technologies have only recently proliferated. Consequently, most patients, clerks, and referring physicians currently do not have access to the information that they would need to schedule patients into surgeons’ clinics based on the surgeons’ OR queues. Instead, patients are usually referred to surgeons and/or choose surgeons based on the surgeons’ expertise and the availability of surgical clinic appointments.
Prevention of oscillations requires EWPS to prevent clerks, physicians, and/or patients from looking ahead at surgeons’ OR queues when scheduling patients into surgeons’ clinics. This can be done, for example, by simply not giving the person who is scheduling a surgical clinic convenient simultaneous access to information on surgeons’ OR queues. Only once the decision has been made for the patient to be scheduled for surgery should a clerk evaluate the surgeons’ OR queues and appropriately schedule the patient into available OR time.
Hospitals and physician practices are now implementing EWPS. If a health care system does not address this problem during implementation, unwanted oscillations in OR workload will start to occur if information about OR queues is used by clerks when scheduling patients into surgical clinics. Consequently, our recommendation is that health care systems program the EWPS to ensure that patients continue to be referred/scheduled into surgeons’ clinics based on surgeons’ expertise and availability of clinic appointments.
There may be two other methods to prevent the unwanted oscillations in OR workload from occurring. First, rules can be used to specify the method of choosing to which particular surgeon a patient is referred to when the referral could be made to any one of several surgeons. For example, a surgical group may decide to assign patients referred to the group to surgeons in the manner that maximizes utilization of the surgeons’ time in their clinics (4). When a patient is referred to the surgical group, the patient would be encouraged to take the earliest available appointment. Then, the surgeon who would care for the patient would be the surgeon who was assigned to work in the clinic on that day. Under this rule, the patient is first assigned a date and time of the appointment and then a surgeon, not a surgeon first and then a time. The patient would continue to be cared for by that surgeon in further clinic and OR appointments (Figure 1).
A second way to prevent disruption of OR scheduling is to change the paradigm used for perioperative scheduling. Most perioperative costs affected by patient scheduling occur in the surgical suite, not surgical clinics (1,12). Consequently, to minimize costs, OR queues would be controlled after patients have decided to be scheduled for surgery (Figure 8), not before (Figure 1) (13). If a patient decides to have surgery, the patient would be encouraged to choose the first available day of surgery. Then, the surgeon in the surgical group who would care for the patient would be the surgeon who was assigned to operate on that day (4). The advantages of this scheduling strategy are that utilization of the surgical clinic and surgical suite will be high and the oscillations will not occur. The (big) drawback is that continuity of care with the surgeon is lost (14). This approach may then be most appropriate for surgical groups and specialties in which the surgeon’s primary role is to perform surgery, as compared with caring for patients during many clinic appointments. This approach would also be limited to surgical groups whereby compensation to each surgeon is not affected adversely by having a different surgeon care for the patient in the surgical suite.
|
These oscillations are easily generated by a number of factors related to the use of EWPS information to schedule patients. Currently available mathematical theory is inadequate to design control theory methods to schedule patients into surgeons’ clinics in a manner that maintains OR efficiency. Our recommendation is that health care systems program these EWPS information systems to ensure that patients are referred to individual surgeons (i.e., scheduled into their surgical clinics) based on surgeons’ expertise and availability of clinic appointments, not the availability of OR time.
|
Footnotes |
|---|
|
References |
|---|
|
TopIntroduction1. Logic Behind Using...2. Pitfalls that May...3. How Using Information...4. Conditions Under Which...5. Review of Management...6. Recommendations in...References |
|---|
This article has been cited by other articles:
|
|
 |
|
  F. Dexter, Y. Xiao, A. J. Dow, M. M. Strader, D. Ho, and R. E. Wachtel Coordination of Appointments for Anesthesia Care Outside of Operating Rooms Using an Enterprise-Wide Scheduling System Anesth. Analg., December 1, 2007; 105(6): 1701 - 1710. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. Moss, Y. Xiao, and S. Zubaidah The Operating Room Charge Nurse: Coordinator and Communicator J. Am. Med. Inform. Assoc., November 1, 2002; 9(90061): S70 - 74. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
