Fighting dysphagia with an unlikely weapon: Math (Part 2)

By George Barnes MS CCC-SLP

Co-author: Doreen Benson MS CCC-SLP

Numbers don’t lie

In part one of this blog post, we focused mostly on the tricks our brains play in altering our critical thinking skills. Now, we are ready to fight those tricks with a little thing called math. In order to effectively manage dysphagia and truly understand the risk of pneumonia, the SLP and the patient need to see what this risk looks like in numbers. Because, unlike your fragile perceptions and assumptions, numbers don’t lie.

Bayes Theorem

“Bayes theorem?? Oh dear, now he's just talking in tongues.” It's ok, take a deep breath. It's a simpler concept than it sounds. Bayes theorem is a form of conditional probability and is used widely in medicine to understand the likelihood of something occurring when there are several factors at play. It is what doctors use to determine the efficacy of certain screening tools and to make a correct medical diagnosis. There are many examples you can use to better understand this concept, but I’ll put it in the context of dysphagia because that’s why we’re all here. 

First, let’s imagine we are seeing a patient and we would like to answer two questions: 

1. Are they aspirating?

2. Will that aspiration cause pneumonia?

Are they aspirating?

We have gotten very good at answering this first question so we don’t need Bayes to help us here. But that doesn’t make it any less important (especially since the second question is irrelevant without the first). While there are ways to make the bedside swallow evaluation more effective, it has been found to be inconsistent across clinicians and ineffective in diagnosing dysphagia. But wait a second... Is that a bird? A plane? No, it’s MBSS and FEES! Instrumental studies, which are currently in a fight for the throne, are both considered to be gold standard swallow evaluations and widely seen as accurate ways to predict aspiration risk. 

Does that aspiration cause pneumonia?

Now that we know the patient’s aspiration risk, we need to determine if that aspiration will lead to pneumonia. Good thing we have some studies to help us out here. Bock et al. answered this very question in 2017 and told us that confirmed aspiration was associated with the increased likelihood of a pulmonary event, BUT only in debilitated patients. Langmore et al. in 1998  and 2002 was more specific and reported the major factors that increase the likelihood of pneumonia. The kicker? Dysphagia was on the bottom of the list and needed to be combined with OTHER risk factors to be counted at all. 

So dysphagia, or even confirmed aspiration for that matter, does not directly cause pneumonia. You think it does, but it doesn’t. This is not to say that dysphagia is not important. It’s still an actor, just not the protagonist. Plus, pneumonia is not our only concern. Dysphagia is a terrible symptom affecting 15 million people in the US alone and about half of people in nursing homes. It can and should be managed properly to improve the lives of the people we serve. But if we are specifically trying to manage the risk of pneumonia, we need to think outside the dysphagia box.

How does one get pneumonia?

Okay Mr. Smarty Pants, if aspiration doesn’t cause pneumonia then what does? Well, I’m glad you asked. While aspiration pneumonia can be tricky to diagnose it is largely dependent on the inhalation of a sufficient volume of pathogens and the inability to clear them out or fight off an impending infection. So, we have the risk for aspiration covered through our instrumental studies, but how do we know if the patient has the ability to clear it or fight it off? Most people simply use their “gut” or they gauge their “comfort” level in order to make a decision. But we already learned that this path of thinking is a slippery slope. 

Back to Bayes:

Bayes theorem helps you take these subjective feelings out of the equation by allowing you to calculate an exact number. The first thing Bayes will call for is a reference point to put the patient’s risk into perspective. This helps us avoid the base rate fallacy by asking, “What are the chances of a patient (any patient) developing pneumonia in this type of setting?” (e.g. at the home, hospital, nursing home, etc.). For example, take a random person from the street who is coughing while drinking bottled water. Like the lady with the glasses who probably wasn’t a librarian, this person is probably not at an elevated risk for pneumonia. Why? Because MOST people are not at an elevated risk for pneumonia. How about somebody over the age of 65? The risk gets higher, but still relatively low (2%). Somebody in a nursing home? Higher (3.7%). Somebody presenting to the ED with acute respiratory symptoms? MUCH higher (15%). This is our reference point or our base rate, and it will vary depending on the population with whom you practice. But this number is not a constant. It will change based on the many moving parts of the person’s physiology and medical status. Everybody will be different. Heck, even the same person can have a different risk from day to day. Such complex problem solving requires Bayes to help you see what you’re looking at.

Bayes allows you to calculate multiple risk factors combined with the accompanying base rate to tell us the probability of developing pneumonia for a specific patient. For example, it will help you answer the question, “What are the chances of an elderly patient developing pneumonia when they have poor oral hygiene, a weak cough after a recent CVA, UTI, and a history of COPD and CHF.” My favorite part? We can separate out the risk factors that 1. Have the largest impact (i.e. debility) and 2. Can actually be managed (e.g. poor mobility vs age) so we know what we should be focusing on. 

Getting out of the vacuum

If it sounds complicated that’s because it is. But the calculation can be simplified with a risk assessment tool. One that’s digital, quick, and easy to use. One that is being researched by the authors at this very moment. Want to know your patient’s risk of pneumonia? Simply answer some basic questions about the patient’s history (takes about 5 min) and what comes out on the other end is an exact probability of pneumonia in the form of a percentage. AND you’ll be able to take out risk factors based on the risks you can manage effectively (i.e. oral care, mobility, etc.) to see how much that probability can be reduced for the patient. It’s time to stop making decisions in a vacuum with aspiration tunnel vision. Maybe then people will finally stop asking us, “Did he pass?” and start asking us, “What’s the risk of advancing this patient’s diet?” And, “What can we do to manage that risk?” 

The best tools may not make a great carpenter. But a great carpenter always uses the best tools. 

Subscribe to my blog if you’re interested in hearing more about the progress we are making on this risk assessment tool. Interested in getting involved in its research? Reach out to me directly: George@FEESibleSwallowSolutions.com

Liked it? Why not share it?

About the co-author:

Doreen Benson, MS CCC/SLP is a Speech-Language Pathologist with over 25 years of experience. She is currently employed at Shenandoah Memorial Hospital in Woodstock, Virginia where she pursues her passion for evidence-based clinical practice in evaluation, treatment, and program planning for adults with dysphagia. She has presented at a number of state and national conventions.