Hospital Navigator Solutions

During the past three years our group has been working with a group of NHS district general and teaching hospitals, exploring ways in which OR methods, and discrete event simulation in particular, might help an acute sector hospital to improve its operational performance. Our initial studies focused on pathways for emergency care. In our studies of the workings of four A&E departments and of the use of emergency beds in medical and surgical rapid assessment units and in the hospital as a whole we were able to establish that one of the most important constraints on the ability of an A&E department to meet the national 4 hour waiting time target is the ready availability of beds into which to admit emergency patients. This fundamental problem of bed availability arises from the difficulty that many hospitals have in achieving a proper balance between emergency and elective activities. Looking at the long term, hospitals generally do not quantify the capacity that they must set aside for emergencies in a statistically appropriate manner. Even where this capacity issue is understood, there is no convenient means of “ring fencing” beds specifically for emergency use, except through a physical separation of emergency from elective functions, not always practicable nor desirable. Inaccurate quantification of the emergency requirement can lead to inaccurate estimates of the true elective capacity so that the hospital agrees to what may be overly optimistic contracts for elective surgery with purchasers. Demanding elective contracts create undue pressure to admit elective cases, eroding emergency capacity, and adding to those pressures that arise from the constant requirement to downsize the overall bed capacity.

 

This emergency/elective conflict is exacerbated by some other factors that work in the short term. Intermittently, there are exceptional surges in emergency demand that force late cancellation of elective admissions. Prompt rearrangement of the cancelled patients (a statutory requirement) is logistically complex and time-consuming, with inevitable ripple effects on other admissions already planned. Conflicts in the allocation of bed resource also arise because surgeons, when they plan elective admissions, do so in isolation, unaware of how their planning decisions may conflict with those made by their colleagues working in parallel.

At the level of daily operations, the management of bed availability is one of day long, non-stop, frenetic and stressful activity, largely focused on admitting patients from the A&E department before the 4 hour deadline.

The hospital’s corporate information systems, and, in particular, the Patient Administration System (PAS), do not provide effective support to the resolution of what is, at root, an information problem. Thus, convinced of the potential value of business modelling techniques, one of our client hospitals (Brighton and Sussex University Hospitals NHS Trust) commissioned us to develop a solution to this problem of emergency/elective admissions planning based on the use of discrete event simulation technologies. We proposed a two phase sequential development, the second phase extending functions incorporated within the first:

In our earlier work with the emergency care pathway, we were able to demonstrate that hospital conditions are more accurately represented by a simulation world view that differs from the conventional factory based world view[2]. In the new world view, expressed through use of a so-called Medical Process template working in Arena, resource allocation is mediated through the use of skill sets and operating priority and is guided by complex drivers that encompass ideas involving the ability of the healthcare professional to perform the task in hand and even handedness towards a widely varying patient demand. This approach has proved particularly suited to the extended sphere of study reported here, enabling us to manage the problems of the relative clinical priority of different procedures and of the varying operative capabilities of the surgical personnel available for particular theatre sessions.

 

In developing the Medical Process template, our intention is to create a productivity tool for Arena software users working in the hospital domain. In this project, we have expanded the capabilities of the template by adding ward and room modules which can be dragged from the Arena template panel and dropped into the model window. The ward and room modules support a rich functionality. The user is able to define a ward in terms of a series of rooms. By means of a dialogue box, the user specifies the number of beds in each room; the opening and closing schedules for each room; and the properties of rooms in respect of they deal with patient gender – can a room contain males and females at the same time, or can it contain only one gender ever, or alternate between genders according to demand? The ward module supports the complex logic that guides male and female flows and that optimises the distribution of male and female patients within a ward so as to maximize the opportunity for admitting another patient of either gender.

A considerable complexity arose in the preparation of data for input to the simulation. Patients for elective surgery are held in a waiting list database maintained on the hospital PAS. Unfortunately, the hospital does not insist on the disciplined use of unique descriptors for each of the 1000 or so procedures that its surgeons can undertake. Surgeons enter their intentions onto a card, using varied descriptions. The contents of the card are transcribed into the PAS by a clerk. Many transcription errors occur. Thus, 13000 waiting list entries for urology fall into 4000 separate descriptions relating to little more than 100 actual separate procedures. Furthermore, many of the waiting list entries relate to multiple procedures to be performed during the same overall operation. Sometimes the time required for the individual procedures may be summed, at other times not. We have developed a translation tool to automate the conversion of a waiting list download from PAS into a list of unique, clearly defined descriptors.

 

---