Surgical Manoeuvre: modelling can help a busy day-care centre plan operations more effectively

A complex daily surgery schedule must balance the needs of individual patients with available equipment and recovery beds, and will benefit from employing a planning tool using algorithms that can be accessed through user-friendly software

An ageing population and increasing demands for surgery make it essential for clinics under pressure to use their resources efficiently when planning operations. But many variables and significant differences between patients can make this a complex task for surgeons, nursing staff and planners. Brecht Cardoen and Erik Demeulemeester set out to examine how a decision support system using optimization algorithms can help medical staff improve their scheduling.

“A Decision Support System for Surgery Sequencing at UZ Leuven’s Day-Care Department” describes an exercise to improve outpatient surgical schedules at the UZ Leuven Campus Gasthuisberg in Belgium. A 10-day case study to illustrate the performance of the modelling system that compared the hospital’s schedules with those suggested by the software indicated that it increased the probability of obtaining feasible and better schedules.

Patients: Popularity of Day Surgery

A shift is underway from inpatient to outpatient surgery, whereby patients are admitted and discharged from the hospital on the same working day. Day surgery has become more common because of greater surgical expertise and new medication, is popular with patients, and offers hospitals and healthcare funders advantages. Patients spend less time on the ward, recover in their own homes and are less exposed to last-minute cancellations. Day surgery is also less stressful, especially for children, and reduces the risk of infection from sicker patients. Hospitals can also manage their schedules more efficiently as day-surgery procedures are shorter and standardized, reducing uncertainty and making healthcare more cost-effective.

Priorities: Planning Constraints

At the busy day-care centre of UZ Leuven Campus Gasthuisberg the 13,000 outpatient operations per year account for 15,000 hours of surgery. Patients arrive just an hour before surgery is due to start and, after their operations, are admitted to the post-anaesthesia care unit before being transferred to recovery wards with either beds or chairs until they can go home.

The procedure used to schedule between 45 to 70 daily operations at the centre has two steps:

  • First, patients and surgeons discuss potential days and time slots in which operating rooms are available, which are then assigned according to the patient’s preferences.
  • Then, a day in advance, the sequence of operations is drawn up and patients are told what time to arrive – a system that significantly reduces no-shows.

Decisions about sequencing respond to priorities that reflect patient needs and staff levels. For example, they aim to ensure children are dealt with as early as possible, that patients who may have to travel long distances are given certain operating times, and that peak bed use in the post-anaesthesia care units is minimized to ensure a level workload for the nurses.

Schedules also reflect limits posed by factors such as:

  • bed capacity in the post-anaesthesia care units
  • whether medical equipment is available and the need to sterilize it after surgery
  • and additional cleaning requirements for MRSA-infected patients.

Sequencing: Computerized Schedules

Surgeons and the head nurse determine through negotiations the sequence in which operations are carried out, but although a surgeon may specify a preferred order, the head nurse may change this to resolve clashes between slots. This makes such an approach time-consuming, and the head nurse’s changes may be perceived as unfair or determined by rule of thumb in a way that does not account for variables such as the demand for recovery beds. The researchers wanted to examine whether theoretical operating-room sequencing algorithms could be applied in practice, reasoning that a decision support system generating less-subjective computerized schedules would help. They applied mixed integer linear programming to the problem then created a user-friendly software interface to make it visibly easy for users to understand and employ.

Data was retrieved from the hospital information system or collected manually. The master surgery schedule, for example, specifies the type and availability of operating rooms; medical equipment data lists reusable types of instrument and how long it takes to sterilize them; and different types of surgery have different expected operating times and recuperation periods in post-anaesthesia care unit beds. Information was also incorporated about patients and potential constraints such as their age, a request for a private bed, MRSA, type of anaesthesia and travel distance to the centre.

Testing: a Valuable Tool

The researchers tested the modelling system using data from two weeks in March 2008 in which between 44 and 64 patients a day were spread over eight operating rooms.

They compared 10 schedules provided by the hospital with those suggested by the decision support system, and identified three possible scenarios:

  • The original schedule was feasible: Only one of the 10 original schedules was actually feasible, and while this performed similarly to the optimal schedule, the latter identified a major improvement in reducing the peak use of recovery beds.

  • The original schedule was not feasible, but a feasible schedule did exist: In six instances in which the original surgery schedule would not work, changing the sequence of operations within each slot would generate a feasible schedule. Using the modelling system, the research team identified an optimal schedule wherever there was a conflict, for example, between the availability of medical instruments and beds.

  • The original schedule was not feasible, but no feasible solution existed: In three cases there was no feasible solution to scheduling problems, and the original schedule that the human planner had come up with was unfeasible. These instances were identified by the decision support system within one second, and often reflected complex problems. In one case, for example, the slot beginning at 7.45am only comprised patients who needed an X-ray during surgery, but this service was only available from 9am.

Where no feasible schedule could be obtained, the system proved to be a valuable instrument for testing changes, such as assigning operations to alternative slots or changing the slot starting times, and demonstrated its worth in enabling whoever is planning the schedule to identify viable solutions. To date, schedule planning has focused largely on the availability of medical equipment, and when a problem is encountered, either the head nurse tries to acquire the necessary equipment from inpatient operating rooms or, instruments are cleaned by hand to speed up sterilization. Problems also arise with recovery beds, and the planner does not currently adapt the schedule to account for their limited availability, causing congestion in the post-anaesthesia care units.

Strenghts: Getting the Best out of the System

The UZ Leuven test confirmed some of the strengths of the decision support system:

  • users can direct the purchase of equipment and justify spending;
  • its user-friendly interface enables users to visualize the scheduling process, in turn allowing them to understand the application and its results;
  • heads of the centre gain time from using it;
  • it can be used to test and compare alternative schedules.

The test also identified improvements that would enhance the use of the modelling system. The system is able to determine the best sequence of operations based on surgeons’ estimates of how long these last, but as these are based on the characteristics of patients they can be highly variable – suggesting the need for a more detailed system to code surgery types. Surgeons also often underestimate operation times because of UZ Leuven’s role as a teaching hospital, as trainees take longer to perform surgery – a phenomenon that could also be addressed with improved coding. Linking the decision support system with the hospital’s electronic patient files would enable the inclusion of data into models to be fully automated. As the day-care centre already uses a tracking tool to monitor the progress of operations during the day, this could also be linked to the system to enable real-time scheduling identifying conflicts – and solutions – as they occur.

Related Article

Cardoen B. Demeulemeester E. A decision Support System for Surgery Sequencing at UZ Leuven's Day-Care Department. International Journal of Information Technology & Decision Making. 2011, 10(2), pp 1-16.

& Rankings

Equis Association of MBAs AACSB Financial Times