More thoughts on shift handover from Tom Campion in the US who looks like he’s been real busy:

..For the first part of the summer I took a class. Since then I’ve been working on research. My research involves the handoff process that physicians go through at the end of their shifts. The literature about handoffs says that processes are not very standardized in hospitals, health systems, or on a national level. Vanderbilt University Hospital has done a pretty good job for the past eight years having a tool in one of the primary information systems where physicians (and sometimes nurse practitioners) maintain free text descriptions of a “case summary” and things “to check” for each patient in their care. This data is then printed out and used as a document of reference for when (and, sometimes, if) an outgoing physician discusses patient matters with an incoming physician. Physicians call this part of the process “sign-out,” and the point of it is to prevent important information from falling through the cracks. Although the tool used at Vandy does a good job, it can be improved so that less information falls through the cracks.

So far I’ve created a web-based version of the tool to replace the old version inside of the primary information system. The new tool resembles the old tool in appearance and functionality. The advantage of a web-based tool over the legacy application is I can add functionality easily and users can access the tool in more ways than one in the future. The new sign-out tool can also share data more freely with other web-based tools in production. Of particular interest in sharing are the “to check” items that physicians record in sign-out.

Because the current “to check” section is a blob of text, the multiple things to actually check are not discrete. If they are made discrete, then from a data perspective they can be treated as separate items, and you can do a whole lot with these items. At this early stage in the project, we’re thinking that most to check items are arranged in a event-time-action manner; that is, “check [some event] at [a certain time] and then [take an appropriate action].” That’s kind of how many of the current to check items are recorded at present in free text. I’m looking to develop a structured way for users to enter this data. This way, to check items will be more explicitly defined, which is a boon for all clinicians. Additionally, a discrete element like time can be used to chronologically sort to check items. To check items recorded at sign-out can also be represented in other information systems, like an electronic whiteboard. The event-time-action orientation of the to check item is pretty much a guess at this point, and things could change, especially for the needs of a specific unit in the hospital.

For my master’s I’m going to attempt to establish that structured data capture of certain to check items is better than the free text method currently in use. I will then create and implement the software enhancements to the web-based sign-out tool to perform structured entry of to check items. I’ll probably pilot this on a given unit to see if people use the structured to checks, and assess if they use structured to check (man, I hope so) and what made them use structured to check (i.e. in what ways does it help them perform the task of sign-out better). and then try it on other units and see what level of customization has to be done for physicians in other specialties to use structured to check in the new tool.

Initially my advisor and I carved out a master’s project where we would automatically create structured to check items from current patient data to try to improve pediatric antbiotic dosing, but a senior faculty member put the brakes on that because the project was too ambitious in scope. For a project about sign-out, the focus has to be on sign-out, not antibiotic management or reminders, he said. And he was right. The reminders project is a huge thing and I might focus on that later. We’ll see. Right now I want to complete my master’s thesis by next August and then move on to dissertation stuff.

I’ve learned to program this summer, which has been pretty cool. The two classes using C++ that I took over the past academic year were very helpful, but this summer is when I really learned to program because I was involved in a project every day. It’s basically the same thing they tell you in class. The language I’ve been using is PHP. I like it. It’s not as tedious as C++, but learning about memory management and objects in C++ really helped prepare me for this. Looking back at the start of the summer when I began to learn PHP compared to now being comfortable with the language, there are, of course, a million things I would have wanted to do differently. I’m learning. It’s great.

The sign-out web application has been implemented and physicians now use it housewide. It’s pretty cool to look at log files and see record of people using something I created. Having created the tool means I also have to fix things that break, and I spent a large part of Friday night and Sunday making sure users could print. The process of understanding a problem, writing code, testing it, implementing it, and maintaining it has been very interesting. Tedious at times, but a great experience.

On the topic of learning, graduate school year two started about three weeks ago. The beginning of ’06-’07 is a lot easier than last year. I understand better what’s expected of me in computer science/quantitative coursework, and I feel like I have a better feel for my interest in the field of biomedical informatics and my role as a graduate student. Last year was just complete and utter shock. I still go through swings of “what the hell am I doing with my life?”, and they’ll no doubt continue, but overall I’m much more calm and focused than a year ago. At least this week.

This post has been rambling. The summer has been fun–at times too fun. My funds have been depleted by the confluence of warm weather and bar tabs. I probably could have gotten more work done and saved more money but oh well. I’m in a good place and I think I’ll get more done this fall. Class this fall will be tough. Algorithms is going to be super hard. The other two classes should be manageable but challenging. I need to stay on top of research stuff, determine my hypotheses, push out features to users, and take some measurements.

Back off, man. I’m a scientist.