Recommended Model for QA/QI in Healthcare: Epidemiology and Biostatistics, not PDSA! Part 5 of 5

My most recommended model for QA/QI in healthcare was requested by a customer who had a department full of amazing professionals. They already were excellent at most of the functions of a QA/QI department that I described in my alternative model in the previous post in the series.

• For the surveillance function, they were already running many successful monitoring systems.
• They also were very good at keeping documentation about anything they did – PDSA stages, the evidence and process they used when prioritizing issues, actions they did to introduce changes, and attempts they made to benchmark changes.

But all of these functions were really hampered by their lack of understanding about study design and basic statistics. So the only function they were really having any problems with was the research function.

My Recommended Model for QA/QI in Healthcare: Empower Professionals with Research Skills

I think this department was like most QA/QI departments, in that their employees know more than I do about healthcare QA/QI because they do it all day. But unfortunately, there usually aren’t straight up research professionals like me in those departments. Most QA/QI departments are filled with clinicians (nurses, physicians, physician assistants) who have been cross-trained in other fields, like research.

But you really need someone like me – who does not do clinical care, and only does research – to be able to mentor a QA/QI professional. This is because it requires years of experience in applied research to be able to give good recommendations of how to approach research study designs in a specific situation – such as a particular healthcare setting.

Here are a few examples of how this department struggled to do research:

• If they needed to make a survey, they did not know basic survey design, so their survey wouldn’t get them the data they needed to answer their research aims, and sometimes had odd items on it that did not make much sense to me. Although they might make it to the point of actually collecting data, they would run into trouble somewhere in the process of actually getting the data and analyzing it to answer their research aims.
• If they needed to use data from a system to make a benchmark, since they didn’t know basic public health informatics, they were lost as to how to structure the process. Even if the IT department stepped in to help, they really didn’t have the subject matter expertise needed to help the QA/QI professionals pick variables from the data they had available in the systems. The QA/QI professionals often didn’t even know you needed to pick variables, and once I explained it, there were still not exactly sure how to do it. They would run into trouble trying to make a plan to analyze the data to answer their questions.

So this illustrated that the only problems they were having were in the research function – they were pros at everything else. So unlike PDSA training, which introduces a completely novel approach from outside healthcare, the training I do on my model just focuses on empowering QA/QI professionals with practical research skills that are used throughout healthcare and public health, and are accepted as levels of evidence in the peer-reviewed literature. That means that if you do a really good case study – like Intermountain Health did – you can publish it in the peer-reviewed literature.

Training using my recommended model for QA/QI in healthcare is done in three steps. These steps can be entirely self-study (totally painful but worth it), or else, I can physically come to your organization and train your QA/QI department in person (way more fun!). But there are also a lot of resources that can help you in between, and I’ll link to some of them in the descriptions of the three steps to my QA/QI model below.

Step 1: Learn (or Refresh Knowledge on) Basic Study Design

Unlike most engineers, who were the targets of the original theory on which the PDSA was based, people who work in QA/QI departments generally have advanced degrees, such as clinical degrees, or master’s degrees in administration or public health (like me!). That means they have already been exposed to basic study design and statistics, either though courses, or by simply reading the scientific literature which they have to do in order to come up with solutions to quality problems.

So the first step to having your QA/QI department do the alternative model is to make sure they understand basic study design using the same terminology used by the Centre for Evidence-based Medicine as well as epidemiologists. Here are a few resources for this first step:

1. LinkedIn Learning Courses in Study Design: Take my online LinkedIn Learning course series in study design
2. Other Online Resources: There are many excellent online resources in the public domain, like this one from Boston University. I also have some useful lectures on my YouTube channel.

Step 2: Ensure a Basic Grasp on Statistics

This one is a little more straightforward, because there are a lot of resources for learning statistics online.

1. Easier: The main one I would recommend is my series of lectures on basic statistics that I originally made for an undergraduate nursing course. It was assembled into one big huge 8-hour lecture thanks to the wonderful people over at Free Code Camp. The slides to the lectures are available here.
2. Harder: The Free Code Camp lectures focus on univariate and bivariate statistics, and don’t explore categorical outcomes. If you want to get your multivariate on, try my courses on R and SAS on LinkedIn Learning.

Step 3: The Next Time You Need to Do Research, Have an Expert Mentor You in Writing a Research Protocol

So once the QA/QI professional feels comfortable with basic study design and biostatistics, the next step is to have them do some research – but of course, you have to wait for an opportunity. In the example I used in my last post in this series, we pretended we were the head of QA/QI of a healthcare system, selected a health literacy tool, and measured a sample of our patient population. That’s a good example of a research study that would need to be done.

Once an actual research study is determined to be needed, the QA/QI professional can get someone who is an expert in study design (like me!) to mentor them through developing a research protocol to complete this work. The mentor should have a lot of experience and a proven track record (like me!), so they can show the professional how to develop a protocol so it is clear, covers all the necessary topics, and makes it through their ethics board or Institutional Review Board (IRB) easily.

Then, the mentor should meet regularly with the professional as they complete the tasks outlined in the research protocol. That way, the mentor can help troubleshoot if anything goes wrong during the study execution, and also provide guidance on the final statistical analysis and interpretation.

So my recommended model for QA/QI in healthcare is both simple and hard at the same time. It’s simple in that it involves training already highly-trained, intelligent professionals in some more skills – which are research design and basic statistics. And these people are not blank slates – many have received similar training before. But what’s hard is that the training is applied. We do it in the QA/QI leader’s context. At that means not only learning how to design studies and understand basic statistics, it means learning how to write research protocols and understand the scientific literature.

The reason why my model always works is because epidemiology and biostatistics always work if you are able to do them properly. On the other hand, the PDSA has not really worked for my customers, and does not have an evidence base behind it. Therefore, I strongly recommend my alternative model, based on the functions of a QA/QI department in healthcare.

Updated June 19, 2021 (added slider menu).