In my quest to develop a mathematically-based theoretical framework for understanding the relationship between clinical work and clinical systems, I have been reviewing theory-focused articles in the academic informatics literature. So far, I have waded through socio-technical theory and human factors engineering. Interestingly, neither uses mathematics to convey their points of view. Why is this?
My research began with socio-technical theory because of how frequently it appears in informatics journals. Papers not focused on clinical informatics were also included as background reading. All were very interesting, but ultimately, it was not easy to compare findings or suggestions. In cases where frameworks were offered for the purposes of analyzing implementations or implementation effects, no objective criteria were offered (at least that I could find) that cleanly separated categories or that identified thresholds for the criteria offered. Definitions for terms also tended to be imprecise. This is not a specific critique of socio-technical theory because the same could be said of “data quality,” “high quality care,” or a number of frequently encountered terms.
More recently, human factors research has been my focus. Papers in this area are becoming more frequent, which is a good thing. I have found that, at some point, every theoretical framework that purports to tie clinical work to information system use discusses processes. Unfortunately, process data are presented in many different ways with authors using an assortment of definitions, diagrams, and terms.
As someone interested in clinical care systems and how they interact with real-world processes, I had hoped to gain much more useful information from my literature review than I have. Don’t get me wrong. I am not saying the papers I read offered no information, only that it was not easy to compare across researchers or fields. Here is one example of what I mean.
Workarounds are a hot topic, yet there is no taxonomy of workarounds that everyone uses. And the taxonomies that have been offered are not tied to objective criteria or an algorithm that would tell one how to recognize or categorize a specific workaround. All things considered, I think it is time to use mathematics in describing clinical processes and clinical work.
Processes consist of a series of tasks, and in turn, tasks can be separated into steps. Within any setting, what clinical professionals do can be represented using processes, tasks, and steps. Workflow patterns, which have been around for about 14 years, are capable of capturing not only task sequences, but also information use and resource interactions. Why are they not used in socio-technical, human factors, or other research that addresses clinical work and clinical processes? If clinical informatics, especially those sub-components focused on clinical work and its relation to clinical information systems, is to move forward we need to publish research findings in a form that allows for easy comparison across researchers.
I suppose one major hang up is that nearly everyone equates mathematics with numbers. Mathematics is not about numbers; it is about patterns and precision. Reading Advent of the Algorithm, by David Berlinski, illustrated these traits better than any other book I have encountered. This book details the painstaking work of numerous mathematicians over the centuries to assure that every definition was logically sound, that mathematical descriptions of patterns or methods were always reliable. The rules of logic and arithmetic were hammered out over years. Before we go any further, clinical informatics needs to address the lack of precision in terminology and analytical methods applied to clinical work and clinical information systems.
I am not saying there should be a grand meeting to arrive at a consensus–that is not necessary. Simply requiring that everyone submitting findings about workflows and processes use workflow patterns would be a huge start.
Here is a favorite quote from Thomas Kuhn in The Structure of Scientific Revolutions that describes where research on clinical work and clinical information systems seems to be today.
Being able to take no common body of belief for granted, each writer on physical optics felt forced to build his field anew from its foundations. In doing so, his choice of supporting observation and experiment was relatively free, for there was no standard set of methods or of phenomena that every optical writer felt forced to employ and explain. Under these circumstances the dialog of the resulting books was often directed as much to the members of other schools as it was to nature.
When Isaac Newton published Opticks, he moved the entire field forward by providing a means for researchers to compare their work.
Workflow patterns are based in graph theory, and graphs can be used to describe any discrete process. Graphs are also visual—every flow chart is a graph. Thus, graphs are a gentle way of introducing informaticists who are interested in clinical work and software design to the mathematical tools that can help them in their work. It is time to move away from flow charts to workflow patterns when creating visual process models.
EHR adoption has driven research in clinical work and clinical information systems to a degree never seen before. However, the value of these efforts will not be fully realized until all work can be readily compared, analyzed, and discussed. Yes, using graphs and learning to apply workflow patterns has a learning curve, but once accomplished, the benefits will be huge. Adopting workflow patterns would introduce formal mathematical principles to clinical informatics while making it possible for researchers to compare their efforts.
Now, back to my reading…