Bob Luttman, Robert Luttman & Associates
Halo Effect
Operational Issues: Documentation and Reporting
Traditional Pathway / Variance Documentation
The Halo Effect
In some ways the Halo Effect is related to the So What problem that we discussed last week, and will again this week. It arises when people examine whether LOS (or some other outcome) is better because of the clinical pathway's implementation.
The analysis usually looks like this:
Last year: X days
This year: Y days
If Y < X, the pathway improved LOS.
Some even go one step further and add the confidence intervals:
Last year: X+/- A days
This year: Y+/- B days
If Y+/- B < X+/-A, our pathway is great!
Wrong! (maybe)
Let's face it: every LOS in every hospital for ever LOS is going down. With and without clinical pathways. Look at the following example:
This data represents a real occurrence (though the data has been altered to mask the guilty) involving hip replacement (the "With Pathway" group) and knee replacement populations. The hip pathway was implemented first with the knee to follow based on learning from the hip experience.
As you can see the knee patients' LOS started down the month after the hip pathway was implemented. They caught up within six months. The knee pathway was never implemented.
What happened is the Halo Effect: Process and practice changes implemented for the hip pathway, and the focus the hip pathway brought to Orthopedic Surgery in general, also reduced the knee replacement LOS.
While most populations are not as closely related as this example, the overall focus on reducing acute care hospital LOS is a variable that any LOS analysis must control for.
The simplistic Y<X analysis, even using confidence limits, is not valid and could cause you to assume your pathway is doing well when improvements are warranted.
