Patient Behaviour Rules

The fundamental feature of ABS is that the overall dynamics emerge or are driven by the behaviours of the locally defined agents. The rules described in this section will govern the behaviours of each patient agent and its interactions with other patient agents and with the ward environment.

4.4.1 Patient Admission and Discharge

Patient interarrival times are sampled from a globally defined interarrival distribution and, on arrival, the patient agent joins an artificial queue from which it is admitted to the ward when there is at least one empty bed (excluding isolation beds). If more than one bed is available, the patient will be randomly allocated to one of them. On admission, the patient has a certain probability of being already colonised with MRSA (i.e., primary case) or not (i.e., susceptible patient). However, the patient’s MRSA status is unknown to the ward until the patient is screened and the test result is ready. The vulnerability of the patient is also determined at admission by sampling whether the patient has been or will be admitted to ICU and whether the patient will need an invasive device during the ward stay.

Agent-based Simulation Model of MRSA

Two separate and independent distributions are used to sample patients’ lengths of stay, depending on the patient’s MRSA status on admission. The patient will be discharged from the ward when the sampled length of stay is met regardless of his/her other states in the model.

4.4.2 Detection by Admission Screening (Primary Case)

If admission screening is operational, each patient will have the MRSA screening test within two days of the admission. Assuming 100% test sensitivity and specificity then the MRSA status will be detected following a screening delay (the time taken for the pathology test and the communication of the results). If a patient has a positive result then intervention policies may be applied (see Section 4.3.6). During the screening delay, no interventions are introduced and the patient may transmit the MRSA pathogen to other susceptible patients. If the sampled length of stay is short, it is possible that the patient has already been discharged by the time the screening result is reported.

4.4.3 Acquiring Colonisation

For a susceptible patient, the screening test, either admission or repeat screening, will simply confirm the negative status with no following intervention policies implemented. The most important issue for a susceptible patient is that he/she may become colonised during the ward stay (i.e., secondary transmission). The modelling of the transmission of MRSA is discussed in detail in Section 4.5.

4.4.4 Detection by Repeat Screening (Secondary Case)

When a susceptible patient becomes colonised with MRSA, the positive MRSA status will not be detected immediately and the patient will start to pose a risk to other susceptible patients. The colonisation of previously susceptible patients will only be detected if there is a policy of repeat screening and the results from that screening arrive before the patient is discharged. If the colonisation is detected, the patient may be isolated and/or receive decolonisation treatment.

4.4.5 Decolonisation Treatment

Decolonisation treatment will last for a few days and at the end of the treatment, the patient will be tested, maybe several times, to see if MRSA has been successfully

Agent-based Simulation Model of MRSA

cleared. A patient who is deemed to be successfully decolonised is assumed to have returned to a susceptible state. A patient occupying an isolation bed following successful decolonisation will be transferred to a ward bay if there is a spare bed available; otherwise the patient continues to stay in the isolation bed. If the decolonisation fails, the patient will go through the treatment again. A colonised patient receiving decolonisation treatment is assumed to have a lower infectivity than an undetected colonised patient. Figure 4.2 illustrates the colonisation status, detection and decolonisation states of a patient and the main state transitions.

Figure 4.2 Patient colonisation status, detection and decolonisation states and main state transitions

4.4.6 Patient Location and Movements (including Isolation)

After admission, a patient may be moved from one bay to another. Furthermore, detected colonised patients will move to an isolation bed if there is one available. Not all isolation beds are used for the sole purpose of isolating MRSA colonised patients.

Agent-based Simulation Model of MRSA

parameter representing the probability that an isolation attempt fails due to a reason other than all isolation beds are occupied by MRSA colonised patients. As a result, isolation may fail even when the model suggests that there are empty isolation beds in the isolation rooms. Under such circumstances, to reflect the real ward situation, a random patient in the bay rather than the colonised patient will move to an isolation bed.

A patient who stays in an isolation bed will normally remain in isolation until the patient is discharged; the only exception is that if the decolonisation treatment is found to have been successful then the patient may move out. Figure 4.3 illustrates a patient’s ward locations and movements in a hospital ward with three ward bays and one isolation room.

Figure 4.3 Patient ward locations and movements in the hospital ward 4.4.7 Type of Patient Agent

In the MRSA model, the behaviour rules of patient agents are normally reactive rather than proactive in nature. In other words, patient agents are largely controlled by the hospital and their behaviours are highly confined by the hospital environment. As a result, patient agents in the model are not as intelligent and adaptive as the agents in some ABS models. Among different definitions and categorisations of ABS and agents, the patient agents in the model fit in with North and Macal’s concept of agents (North and Macal 2007). In their definition, agent potentially may have five different properties (see detailed description in Section 1.3.3) and agents may have some but not all of these properties. The important thing to be deemed as ABS is that the model is structured in such a way that missing features can be easily added within the established modelling framework. In the proposed model, patient agents are self-

Agent-based Simulation Model of MRSA

contained independent individuals that are situated in the hospital environment and interact with other agents and the environment. They are not adaptive or proactive, but they are still agents who have complex reactive behaviour rules.