Work Domain Analysis – Health, Technology & Organisation (HTO)

How to separate the trees from the forest – The workplace as a Swiss army knife

Continuing on the “forests and trees” metaphor from a previous post we will now see what happens when you try to separate the trees from each other. So, let us start with one single tree (or task). One tree does not make a forest, but it can very easily be distinguished as a tree of a certain kind. A Pine tree has long needles; A fir tree has short needles. And a tree without needles is a leafy tree (unless it is a Gingko tree, but that is another story). However, once there are more trees in a lump, the categorisation becomes more difficult.

Transferred to the issue of work and tasks: As long as we are only doing one single task or having one single role, it is also quite simple to see and study it, and also (at least to some extent) to understand how it works, and what the consequences are. This is not to say that this kind of work is simple or uncomplicated, a heart transplant is a complicated task, even though it is a single task and well specified at that. We may also still make design mistakes on this single task design, but these kinds of mistakes are mostly confined to the task environment we work in at the moment.

In a way, as long as we have many instances of the same type of single tasks that we can handle more or less sequentially, we are still proceeding through something, which in the forest metaphor is like a well-managed orchard with the same kind of trees planted spaciously and in neat rows. We can still handle the situation with the same kind of specialised tools, and the tools may also have an easier way to communicate with each other, e.g., using the same terminology, the same data formats etc. since they are designed for the same task.

Now, when we consider the work situation for nurses at a hospital ward explicitly, their situation is completely different. Their work more or less constitutes the very definition of “fragmented work tasks”. First, there is planned fragmentation, where the person has to change tasks or responsibilities according to a schedule. Second, there are interrupting tasks in terms of alarms, visiting relatives, telephone calls and other randomly occurring events that also cause a fragmented work situation. There is a related blog post by Åsa Cajander “On Digitalisation and Fragmentation of Time” about this phenomenon.

The interruptions that occur over a work day can have at least two major effects on the nurse. He or she has to switch context for the work, which in itself is a stressing action for the brain. In the case of sudden intruding tasks, there is also the problem of mentally going back to the interrupted task and continue from where it was discontinued, sometimes after many hours. This situation becomes almost, again in the forest analogy, like looking at a primeval forest or a rainforest. There are trees everywhere and of every kind, and they are all interconnected in some ways. You need both stamina and focus in order to find your way through the forest/work.

This becomes even more problematic when we look at the computer support for work tasks in any workplace. The general computer system is in many cases designed in the same way as a Swiss Army knife. We use the same technical framework for all the tasks that should be performed, using a keyboard, a mouse (or a mousepad) and a screen for the stationary tasks, and mobile apps, in the best case redesigned for the smaller screens of mobile devices, for the tasks that are done on the run. Journal entries, medication dosage calculations, and many planning tasks are therefore performed in the same limited cognitive space, and there is no real support for the switching between those tasks, nor for the fragmentation of the ongoing work.

In some cases, like writing down the notes for the daily ward round, the information is even taken down as hand-written jots on a paper form and has to be transferred manually into the journal later.

A Swiss army knife is not really perfect in any of its functions (even the knife blade leaves a lot to be desired), it still works, and someone who is not a carpenter, mechanic or plumber might even think it is a good and practical tool. But for any professional, if asked, they would definitely prefer to use a specialized tool. Apart from being more efficient, specialized tools also pose a lower risk of accidents and slips. In the same way, non-specialized computer tools may, even though they can perform the desired tasks, still not be the best choice in a complex work situation.

When we design new computer systems for organizations like hospital wards, schools, etc. they are often positioned around one task at a time. We study one tree at a time in the forest. Even though we are much better off than at the beginning of the PC era, when we could only run one program at a time, we are still running several parallel processes that interconnect badly, if at all.

During the study visit at Akademiska in Uppsala last year (the university hospital in Uppsala), I found ten different systems used in one of the wards, all running on the “same computer”. This amounts to having ten tools on the Swiss army knife. Of course, the individual programs are in some way specialized. We don’t use spreadsheets for everything. Each separate task has applications or programs that are explicitly designed for its purpose but not for the whole entangled web of work tasks. Furthermore, the interaction design is often completely different in the different systems, which means that the user has to switch from interaction style to interaction style, and from command structure to command structure.

One example to indicate what I mean: the notes that are made for each patient during the ward rounds are almost always written by hand at the ward where I visited. However, after the ward round they have to be entered manually into the system, which has the double effect that it might be a slow process, taking valuable time from the nurses, and that there is also a high risk of misreading numbers and hastily jotted down notes.

The paper notes are not integrated into the computer system. While we note this, we might realize that we could need a more automated solution (which may still need proof-reading and checking). In this new solution, it would be possible for the doctor on duty to write, still by hand, on a tablet that may offer character recognition. This would reduce the tiring process of transcribing the text, making it much easier and also leave more time for other, more important tasks.

Now, this is, in my opinion, where one of the problems with the systems in the hospital resides. We are so focused on the design of the single system, that we forget all the small details that could make the whole work situation simpler and more manageable. Add to that that we also forget to look at the small details that should incorporate such things as:

specially designed portable devices, that are designed for a single task, such as taking notes for the information rounds.
the development of distinct, but consistent interfaces to all the subsystems.
ncorporating support for the interruptions that occur during a day. Why not a “what did I do last”- kind of function?

The conclusion of this observation is that we need to look at a work situation from a holistic and a reductionistic perspective simultaneously when we are planning and designing the digitised work place, resulting in a well planned and easily manageable rain forest of work tasks.

Work Domain Analysis of an Intensive Care Unit: a study on how digitalization affects nurses’ work

During the spring Sara Carlson, from the master program in Human-Computer Interaction at Uppsala University, will work on her master thesis about how changes in a work system due to digitalization can be modelled. She will conduct observations and interviews with nurses at an intensive care unit to investigate the effect of the new system that have been implemented. She will explore how these changes can be modelled in an Abstraction Hierarchy from the Cognitive Work Analysis framework, and compare her results to an analysis of the same intensive care unit made prior the system implementation.

We are looking forward to work with Sara and her interesting study, more updates will come!

What we see and do not see – Some further notes on the observation studies…

Doing observation studies is not always as easy as it may seem at first glance, and Diane has already written much interesting stuff in her previous posts on the topic. I agree with her meta-observations, and I just thought I should add some of my views on this topic as well. My experiences are from my two days of observations on the children’s hospital, and the ward for children with tumour and blood diseases. Although the doctors and nurses work with very serious issues, I only experienced a very constructive atmosphere during my two days.

Most of the time when we visit the hospital environment we are there to receive treatment or visiting someone who is. We see a lot of things, and in some way they make sense. We see the blinking numbers on the wards, and we see the different screens used by the nurses and doctors. We see the white coats with blue stripes and don’t think more about that. But when we set out to actually study what is going on in this environment we not only have to see but also interpret the observations into scenarios. Even when we want to study a single system, and its effects on the work, it quickly becomes very evident that the workspace is a very entangled mesh of interactions between people and people and between people and machines. Some of these interactions are very short but nevertheless less important, and they are easy to miss if you are not focused. Some interactions are longer and thus also easier to observe. But in some cases, the more long-term interactions are also easy to miss because they are not continuous and thus have to be observed not only in terms of the shorter sequence as a whole (for instance, see example 2 below).

So, what do we see during an observation visit? Lots of things, but it soon becomes clear that the things we do not (normally) see, are just as important, if not more. Just to give two quick examples:

1. In the ward, at every workplace there are two screens for logging into applications, such as Cosmic, and other supportive tools. What is not (easily) seen is the parallell information storage that is widely used by the nurses. After almost a day at the ward, I suddenly realised that all the nurses had a small paper notebook, which they consulted now and then. It was kept in the pockets of the coat but was very difficult to observe. When asked about it, the nurse told me that the notebook was used to keep track of the details about each patient. The notebook seems to be an important but almost externally invisible information carrier.

Another observation that caught my attention, not because it was evident, but rather because it was not, was the role played by the alarm bell. The alarm has two functions, one that is an emergency call and the other, which is just a call for help with toilet visits or similar. Both are noted on the same display, and with similar sounds (still clearly easy to distinguish). However, the reactions to the alarms are completely different. In the second case, one or two nurses go over to the room, as soon as they are finished with their current tasks. In the first case, the work spaces are emptied within a few seconds. All tasks are interrupted, and almost everyone rushes to the room in question. Since they rush in the middle of a task, the software applications need to be extra supportive and help the nurse getting back into what he or she was doing. This is not something that is easily visible but could be of great importance.

These two examples show in a clear way that observations can be multilayered and need to be both seen and put into the work context. In the case of the notebook, it was also something that was not really thought of by the nurses; it was so integrated into their work that they never gave it any thought.

This makes on-site observation studies both important and interesting but also difficult at the same time. How to systematically get at these ”invisible” observations is a difficult matter, and from my experiences, I think it requires a long observation time to find many of them.