Dr. Peter Pronovost recently penned a viewpoint piece for JAMA about how guidelines can be better implemented. He is well respected in the patient safety realm and clearly feels guidelines are a major way to improve patient safety. I agree that they are a piece of the puzzle. What I wanted to do in this post is critique his thoughts on enhancing guideline use by physicians. I think some of what he proposes is unrealistic at best and most likely impossible.

Let’s take one step back to look at barriers to guideline implementation that were identified in an important review by Cabana and colleagues in 1999. They performed a broad systematic review of 120 different surveys investigating 293 potential barriers to physician guideline adherence.  The figure below outlines what was found.

Barriers to Guideline Adherence

With this background let’s analyze Dr. Pronovost’s 5 strategies to increased guideline adherence.

1. Guidelines should include an “unambiguous checklist with interventions linked in time and space“. He recommends key evidence-based practices. I concur with this recommendation. Checklists are something physicians can do and they have been shown to reduce harmful events. Also this would be behaviorally based and specific (i.e. something measurable). An example might be an item on a checklist to make sure that each day an assessment is made in the chart for the need for continued bladder catheterization. What I worry about is that too many recommendations are made in many guidelines. Recommendations need to be prioritized and limited to those things that really make a difference. The checklists could become so burdensome that they impair patient care as too much time will be spent checking off  the check list and not enough time actually caring for the patient.
2. Guideline developers should “help clinicians identify and mitigate barriers to guideline use and share successful implementation strategies“.  Here is the impossible. While I agree in principle with this recommendation it can’t be implemented. Barriers are a local phenomenon; often a hyperlocal phenomenon. My hospital has 3 separate primary care practices (1 a resident practice and 2 fulltime provider practices) with very different types of docs and practice patterns. What will work in one of these clinics won’t work in the other. What works for one practitioner might not work with any other. Large centralized guideline developers just can’t be expected to develop solutions to barriers.
3. Guideline developers could “collaborate to integrate guidelines for conditions that commonly coexist“. Most guidelines are single disease guidelines developed by single specialty groups. Patients are multimorbid. This is a great recommendation that will be tough to implement but that I agree with 100%. Diseases and their treatments interact with each other. Guideline developers often ignore these interactions. At best they will discuss some exceptions to the guidelines for other comorbidities but this isn’t enough. There are enough diabetics with hypertension and coronary artery disease to warrant a guideline on them. What about hypertension with renal disease? The combinations would have to be carefully thought out and the panels multidisciplinary with primary care physicians playing the prominent leadership role.
4. Rely on systems rather than the actions of individual clinicians. Bravo. Many things are not totally under the physician’s control and there are often too many things to think about now a days for 1 person. Systems need to be engineered to deal with the mundane things we physicians don’t like to deal with (like elevating head of the bed in a ventilated patient. We would rather deal with managing the ventilator). Multidisciplinary teams at each care site would need to be put together to design the processes of care.
5. Create transdisciplinary teams to develop scholarly guidelines with practice strategies. Not much detail is given in the manuscript about this but what I think he means could be twofold: 1) teams of clinicians, epidemiologists, implementation scientists and systems engineers would develop the guideline and 2) these same types of teams would study best practices for implementing them. Currently many guidelines don’t include implementation scientists nor systems engineers. It’s no wonder we have a hard time implementing guidelines with implementation isn’t really built into them from the start.

Much of this is already known but important to keep saying. Someday guideline developers and policy wonks will listen. Just shoving a guideline in our face isn’t the way to go. Currently reminders and “performance measures” are the main ways guidelines are being implemented. We will see if medical systems develop smart ways to use electronic health records to better implement guidelines.