Resource capacity allocation of inpatient clinics AMC

An ageing population, more advanced treatments and a high standard of care led the past decade to an enormous increase in demand for care and costs. Health care managers face the challenging task to organize their processes more effectively and efficiently [17]. Within the Academic Medical Centre of Amsterdam (AMC) the sense of urgency to change is gradually accepted. Different types of research projects are started in order to improve the overall performance and to provide insight in the relations of complex hospital processes. High fluctuations in the demand for care and beds in the clinical wards of the surgical division of the AMC have led to the development of two models. The model of Smeenk et al. [29] makes it possible to predict the number of beds that are occupied each hour of the day given the Master Surgical Schedule (MSS). The model of Burger et al. [9] uses the output of the model of Smeenk to determine the optimal number of dedicated nurses per ward and the number of nurses per flex pool. A flex pool consists of nurses that still need to be assigned to a ward at the start of a shift given the dedicated nurses already assigned and the number of patients present. The models of Smeenk et al. and Burger et al. focus on the clinical wards while the MSS is created in the OR department. In this research we develop an integral method that encompasses resource capacity planning decisions in the OR department and the clinical nursing wards. We have formulated the following research objective: To develop a method which determines the best combination of patient case mix, OR capacity, care unit and nurse staffing decisions in such way that total cost margins are maximised while satisfying production agreements and resource, capacity, and quality constraints. We express our research objective as a mathematical optimisation problem in which we minimise the resource usage in the OR department and clinical wards, while selecting the most profitable case mix. We define several quality and resource constraints. To evaluate the total costs of the objective function we have defined several cost parameters. The solution method we present encompasses a decomposition approach in which we use several models and optimisation tools based on state of the art literature. Our solution approach consist of the following six steps: 1. Set the desired patient case mix and the length of the MSS. 2. Solve an Integer Linear Program (ILP) to create a master surgical schedule and assign elective and acute patient types to wards, while minimising the number of ORs, wards, and the expected number of nurses and beds required. 3. Evaluate the access time service level of the created block schedule with the model of Kortbeek et al. [19]. 4. Determine the number of beds required per ward while satisfying target rejection and misplacement rates with the model of Smeenk et al. [29]. 5. Iteratively use the model of Burger et al. (Step 6) to determine the best flex pool-ward combination. i 6. Determine the optimal number of dedicated nurses per ward and the total number of nurses in a flex pool given various target service levels with the model of Burger et al. [9]. To test our approach we performed experiments with real data obtained from the surgical division within the AMC. Our experiments show that our solution approach reduces variation in demand for beds and thereby levels the workload. When we consider a cyclic MSS of four weeks we can reduce the number of beds by 5.2% compared to our model representation of the current situation. From our results we conclude that nurses can be utilised more efficiently by considering less wards with more beds per ward. When we consider three wards with at most 50 beds we require 11.1% less FTE nurses compared to our model representation of the current situation. When we consider a flex pool of nurses between two wards we can achieve an additional reduction of 1.7% in FTE compared to our model representation of the current situation. The benefits of a flex pool mainly depend on how the MSS is organised, the flex pool-ward assignment and the chosen values of the service levels. Our solution approach encompasses a large variety of resource capacity planning decisions that are related to each other. Due to the large number of planning decisions and the complexity between them it is very ambitious to find one optimal solution. The MSS that results from solving our ILP does reduce the expected number of beds and thereby reduces variation in demand for care. Possible improvements lie in the development of an MSS that further improves alignment in demand for beds with the required number of nurses and a tool to automatically select the optimal case mix. The patient-to-ward assignment can be improved by taking the surgery, and, admission and discharge distributions into account. To conclude, the approach we present provides hospital managers with a tool to evaluate and optimise the resource requirements in the OR department and the clinical wards given a patient case mix and the length of the MSS. This tool can be used to (re)design, evaluate and improve current hospital processes and is, due to its generic nature, applicable in a wide variety of hospitals.