One of the best tips I’ve ever learned about data interpretation is to start big, and work your way down. It is much easier to start with a macro understanding of something, and then to peel away details, then to build up an understanding from a lot of micro observations. For example, looking at all of your individual clients in isolation, and then building that into a report on your program as a whole is not a very natural process of reasoning. Easier, is to first understand your program, and then deep dive into any individual to see how that general and the specific interact in that person’s case. Because of this logic, the first report that we stress in our TCOM workstation, is the Total System Outcomes (TSO) report. Here is a screen shot of the report:
The report is built on the logic of the Average Impact report, as specified in the Praed Reports Suite 2.0, in which the columns represent the average of the count of actionable needs (e.g. items at 2 or 3) in a program. In those original specifications, the Average Impact report only looks at one program at a time. In the Total System Outcomes report, however, we put all the programs in our TCOM system next to each other, allowing us to get one of the most comprehensive, large scale overviews of our TCOM system possible.
For instance, in our graph above, we see on the X Axis each of the programs in our TCOM implementation: Intensive Family Therapy, Community Based, Psychiatric Services, Outpatient, and Truancy. The first column (the green one) represents the average actionable needs upon entering that level of care, and the second column means (in this particular filtering) the average most recent actionable needs score. (By the way, though you can’t see it in this image, if you float your mouse over any of the columns, you are told the count of clients represented in each program, as well as the average length of time between the first scoring and the second one that makes the black bar.)
From this vantage point, you are able to see some of the most important trends in your system. First, you can see which of your programs have the highest needs cases, and which have the lowest needs cases. There are many times that at first review, stakeholders are surprised to find that what they thought was their lowest needs program is actually in the middle of the pack, or some other oddity. This can often lead to quality initiatives, where people are more appropriately sorted to programs designed to deal with the needs they present with.
Second, you want to see the “stairs” effect that you see above: where the program most to the left has the highest needs coming in, and then the score upon leaving the program should be about where the next program’s initial scores start. As such, you are having a system of care that can cover people from the highest needs, and work them down and out of your system. In my experience, you are often hoping to get at least an improvement from the green to the black bar of 1—that would be a whole actionable need improvement for a program. If you cannot accomplish that much, then the program is probably not having a huge clinical impact (though there may be other benefits, that we can see in other reports that I will highlight in another blog post). Similarly, accomplishing at least one actionable need improvement is a clear meaningful impact that can make sense at the highest level to explain what the program is doing in your TCOM system.
The last high level thing that this report quickly shows, is significant “gaps” between the discharge status of one program and the incoming scores of another. For instance, in our example above, notice that the black bar of outpatient is almost a whole actionable need higher than the green bar for truancy. This means that, on average, people leaving outpatient are significantly more needy than what truancy can handle. Now, maybe in this case, truancy is a specialty program and not necessarily a step down out of outpatient. However, that means that people are leaving outpatient with about 4 actionable needs … what do we have for them at that point, because obviously their needs for support have not ended?
These are the kinds of questions that guide you deeper into your analyses of your system. And I hope to show some more highlights of this process in upcoming blog posts.