Prospective risk analysis – a little more “localised”

Although I am a little bit slow in writing this, I’ve recently published some work I did on communication patterns within complex organisations. With a specific requirement that I use agent-based modelling, and the aim of improving our understanding of how and where adverse events happen, I had a look at how hospitals go about doing risk analysis – both retrospective (like root cause analysis) and prospective (like fault tree analysis and a bunch of other things). The major problem with doing risk analysis in hospitals is that most, if not all, of the processes (like doing surgery, caring for a patient, transferring a patient, etc.) involve lots of people, and often, more than one system of information.

At the very least, hospital information systems include chit-chats in the corridors, phone calls, paper-based forms, IT management of medications. In a typical hospital there may be tens or hundreds of systems for communicating information. And with the complexity of work patterns (every patient and every work day is different), it makes the whole thing really tough to model.

AG Dunn, M-S Ong, JI Westbrook, F Magrabi, E Coiera, W Wobcke (2011) A simulation framework for mapping risks in clinical processes: the case of in-patient transfersJournal of the American Medical Informatics Association, 18:259-266.

In the article I wrote alongside a relatively long list of my colleagues in CHI, AIHI and CSE (all in UNSW), I simply came up with a way of translating work patterns into an agent-based model. In this case, an agent is a bit of programming that is designed to represent a person (or an IT system) that has its own goals and a bunch of ways they can use to reach their goals. Kind of like a real person.

What I showed was that even a simple hospital process (like transferring a patient to radiology for a test/scan) has become terribly complex as a consequence of diffusion of responsibility, where lots of people are responsible for the same thing and they all assume someone else has done it. The technical bit of the work is that my approach allows us to map out all of the little mistakes and errors (like not reading a form properly, or not washing your hands) might combine in different ways to create the opportunity for an adverse event – like spreading infections, performing an operation on the wrong person, the wrong body part, or performing the wrong operation all together.

The thing is – we all make lots of mistakes every day – but the organisational processes around us should be robust to these mistakes. When things get complicated, it becomes more difficult to catch and mitigate our little mistakes and they tend to propagate further, and more quickly.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s