Do You Make These 5 Mistakes With Your Quality Improvement Program?


by:  David Kashmer (@DavidKashmer)


Here are 5 of the most common barriers I’ve seen to effective quality improvement in healthcare…


(1) Respond right away to the worst (or best) case

This one is a classic, especially in healthcare.  Some cases feel so bad (or so good) that we re-orient an entire system around that spectacular case.  The problem?  Responding to best or worst cases, without knowing how they fit into your overall system’s performance, actually tends to introduce more variation (and lower quality) into your system!  Before responding to a case that feels terrible (or great) make sure you know the distribution of data for your system, and where this case that felt so bad fits.  Remember:  failure to rigorously know how your system performs means that you run a high risk of creating worse outcomes when you make changes based on one case without fitting that case into the context of your system.


(2) Attribute the problem to the people involved (only)

Here’s another healthcare classic:  attributing some outcome to one person or personal failure.  In healthcare, we’re trained to very much focus on our personal performance, which is probably a good thing.  We’re tough on ourselves.  Often, in training young staff, we focus on an illusion where, by force of personal will, the doctor involved could’ve overcome all the friction in a system to achieve a great outcome.  (More on that here.)  We pretend it doesn’t matter that labs weren’t drawn, the equipment wasn’t there, or that the fact it was raining outside caused 200 patients to be present in the Emergency Department so that the system was over-run.  Failure to recognize the contribution of all six causes of special cause variation leads to an overly narrow focus on people to the exclusion of all else.


Guess what?  People are very important.  However, we’ve noticed that, when the system around the people is cleaned up and aligned with the desired outcomes, that the people suddenly look a whole lot better.  Often, I wonder whether the “people issues” like providers fighting in the Emergency Department, other friction between staff, and poor decision making are manifestations of a bad system.  After all, when the system is repaired, a lot less of what used to be called “people issues” come up.


(3) Fail to recognize the true message of your system

Interestingly, every system tells a story.  Data collection and visualization allow the central tendency of the situation to be appreciated.  Just as importantly, the variation seen in the system (which often relates to the risk that one patient will experience a bad outcome) can be readily appreciated by collecting data and visualizing it.


One fact that makes it tougher to recognize how your system performs is the focus on external benchmarks.  Whether it is the regulated nature of healthcare or not, focusing on benchmarks before you’ve improved your system may lead you to accept a worse performance than you could actually achieve.  Consider improving your system and eliminating defects first and then comparing your new achievement to external benchmarks.  No, you can’t stop measuring government and other accrediting body endpoints (nor should you) yet you can focus on minimizing your defects to the lowest possible number before you start to respond to external benchmarks.


Why is so important to get a clear, visual version of how your system performs?  It allows you to make interval improvement.  It allows the team to see how it performs as a whole.  It makes it obvious that, without improvement, the system performs at a certain level that will have some number of bad outcomes or defects.  It can make the team uncomfortable, yet the discomfort is the sign of impending growth.  Consider that, if you don’t have data, you just have an opinion based on the part of the system you see…as you see it.  More on that here.


In healthcare, once we see our true performance we are left with the decision to improve the situation or just to attempt to ignore it.  Casting the choice in this way can make us feel a little uncomfortable, and yet this discomfort often signals that we are about to break through to a new level of performance.


(4) Never figure out whether you’re doing better, worse, or there’s no change

There are some tough situations that exist out there in healthcare process improvement.  One of the most difficult is when the team feels it’s doing better but in fact there’s been no real improvement.  Just as tough is the feeling that there’s been no improvement, despite hard work, yet (in fact) things are improving overall.


These situations point to the importance of statistical process control, because one of the most useful elements of data driven quality improvement is that it has the power to disabuse us of confirmation bias (seeing what we want to see) and other pitfalls of process improvement.  Data driven quality projects allow us to see statistically significant, meaningful improvement or the lack thereof.  We may even see that, despite hard work, we have yet to see improvement.  In short, data driven quality improvement tells us about the performance of the system no matter what our gut may report.


(5) Failure to create a process that uses rigorous data and monitors your progress

In order to use data to decide what to do, we need to be careful to use meaningful data.  Setting up a process that collects meaningful data, supports the team’s needs, and keeps the team together in a non-pejorative manner is key.


Just as important as having useful data is to remember that your quality system needs to perform ongoing maintenance.  That is, once the team has improved some aspect of the program, the quality system needs to look in on that endpoint to make sure it’s still performing over time.  Failure to utilize ongoing surveillance of previous improvements just allows any gains to erode over time.


One nice way to express this sentiment is that quality improvement projects don’t end, they just enter a control phase.  No matter which technique you use, remember to look in on your previous work at certain intervals.  Be sure you check in on previous work so that you can “trim the weeds”.


Do you make any of these 5 classic mistakes with your quality improvement systems?  Don’t feel bad, most programs in healthcare commit at least one of these on a daily basis.  I hope that focusing on the root issues in your system with a rigorous, data-driven process will allow you to see that many quality issues aren’t as simple as a mistake or poor performance by one of your staff.  Meaning quality improvement that avoids these classic pitfalls can help your system progress to creating less defects and rework…then, suddenly, the people involved look a lot better!