Base Rates: Why the SLP should know (just a little bit of) statistics
By George Barnes MS, CCC-SLP, BCS-S
Edited by Allie Mataras MA, CCC-SLP, CBIS
Base rates are a fairly simple concept. But it’s a concept that can have a profound impact on the way we practice and the outcomes we deliver for our patients. I admittedly didn’t fully understand nor utilize the concept of base rates until fairly recently. Maybe that’s because just the mere thought of statistics stirs my brain like the bartender making me a summer cocktail. I’ve posted about math’s relationship to dysphagia management in the past and have even mentioned base rates in a prior post. But I truly believe the concept deserves its own post because once you understand it, it will change your whole perspective on how, why, when, and for whom you manage dysphagia.
So what in the world is a base rate anyway?
I can give you a definition. Really, I can. For example, I can tell you a base rate is the percentage of a certain population that exhibits a specific characteristic. But that won’t mean much to you and that’s OK. In fact, it’s better to describe a base rate in an example. Say, for some bizarre reason, you want to know the chances that an individual is either a librarian or a salesperson. Meet Susan. Susan is a bookworm with large thick framed glasses who likes to wear her hair up and dress in buttoned up cardigans and long wool skirts even in the summer time. Now, would you say Susan is more likely to be a librarian or a saleswoman? Go ahead, take some time to think about it and come up with your answer. A librarian you say? Good guess, but unfortunately completely incorrect. What you’ve done here is you’ve fallen victim to the bias of base-rate neglect. In other words, you have completely ignored the base rate. The base rate in this instance is the number of people in the general population who are salespeople vs librarians which is approximately an 80:1 ratio in the united states. You let the less important information of what Susan looked like distract you from the statistics. No matter what her appearance is, Susan is WAY more likely to be a salesperson.
When does this become relevant to me as a Med SLP?
We are getting there, don’t worry. Just a little bit more about base rates (I’m sorry, I just can’t get enough. It’s a problem… I know). Base rate neglect doesn’t just occur in “fun” parlor trick games like the one I just played on you (Gotcha!). It happens in healthcare too as do many other biases. Picture yourself walking into a patient’s room to do a swallow evaluation. Let’s call the patient Gerald. In your chart review you notice that Gerald was admitted for community-acquired pneumonia. PNEUMONIA. Oh no! The dreaded pulmonary infection that we as SLPs have reoccurring nightmares about (just me?). After your extremely thorough bedside swallow evaluation, you notice that Gerald appears to have a functional swallow. But how could this be? He has PNEUMONIA after all. So just to be safe you recommend thickened liquids until you can get an instrumental study which you’re hoping will be sometime before Christmas. What you’ve done here (and don’t worry, you’re not alone) is you’ve neglected the base rate. In this case, the base rate is the number of aspiration pneumonia cases that are found in the total number of pneumonia cases in the community, which is estimated to be up to 15%. So without any other knowledge about Gerald, we know that the chances of him having aspiration pneumonia is about 15%, or in other words, it is 85% likely that he doesn’t have aspiration pneumonia. So in a very quick cost/benefit analysis we can assume that the costs of thickened liquids may not be worth the benefits before taking a closer look via an instrumental study.
How else can I use base rates?
Base rates can give you a sort of map of the terrain you are navigating when we so often feeling like we are flying blind. You can use base rates for pretty much anything as long as you have good research to back it up. For example, say you are seeing a patient at bedside after a stroke. Let’s call her Melanie. Melanie has hoarse vocal quality at baseline and left sided weakness. MRI shows a right sided cerebellar infarct. She appears to tolerate all PO trials at bedside without signs or symptoms of aspiration. Great- Regular solids thin liquids it is! Now that it’s only 3p I can finally take my lunch break. But wait a second. What about the dreaded silent aspiration? The aspiration that creeps up on your poor patient when they least expect it like a thief in the night. How do we combat something that’s essentially invisible in the darkness? Let’s shine the light of research on that dirty little thief, shall we? It looks like the base rate of silent aspiration for patients status post acute stroke is up to 27%. Not a small number. More than one in every four. Given her changes in vocal quality indicating possible vagus nerve involvement, it may be worth taking a closer look by recommending Melanie for an instrumental study and knowing for sure before we provide a diet recommendation.
We can also use base rates for specific conditions that we don’t have a lot of clinical experience with. For example, by reviewing the research you can find that your patient with a glioma is about 30% likely to have dysphagia (about a 1/3 chance). There’s also a 45% that your patient with dementia in an nursing home may have dysphagia (almost a 50/50 coin flip). You may also find it useful to know that your patient with Parkinson’s Disease may have an 87% chance of having some level of dysphagia. Of course the data will change depending on the severity levels of the condition as well as the presence of other risk factors, but having a base rate is a good place to start. You can always adjust these numbers based on other factors as you dig deeper into the wondrous world of statistics.
What the base rate can’t do
What’s important to remember here is that statistics don’t tell us the future. They can’t tell you what will or won’t happen. Or what is or what isn’t occurring. It simply tells you probabilities based on group research related to your patient. Also, remember that not all research is created equal. Quality will differ and the relevance to your individual patient will differ depending on the study conducted. This is why critiquing the research is so important so that we learn to provide different levels of value to the different articles we find. But in general, getting a base rate is a good place to start our assessment and A LOT better than going in blind.
Watch out for the broken leg
There’s something in statistics called the “broken leg” phenomenon. Say you are trying to predict if a man will go to the movies this upcoming Saturday night (I guess people still go to the movies, right?) You do all the background research you can. How much does he like the movies? What movies are out right now and what does he enjoy watching? You can even get a base rate by determining the percentage of Saturday nights people like him go to the movies in an average year. But then you find out that he recently broke his leg and so attending the movies now becomes MUCH less likely. This trumps all the other information we acquire in our research and pretty much solidifies our prediction that he will not go to the movies. A “broken leg” in dysphagia management might be a seeing a patient who is orally intubated, confirming the presence of gross aspiration on an instrumental study or witnessing persistent, and strenuous coughing on a particular consistency at bedside. Depending on the goal and what question you are trying to answer, this information might trump any research you find on a particular condition.
What does it all mean, Basil?
What this means is that the research and the statistics will only take us so far. They can be extremely helpful in many contexts, but eventually you have to go in the room and evaluate the patient. The data can tell you how likely your patient is to have dysphagia, aspirate, or develop aspiration pneumonia and THAT may help you determine if it’s worth doing extra testing, making additional referrals, or providing a plan of care that is more or less conservative. I think I like base rates so much because they give us a nice, neat picture of what might be going on with a patient before we even walk into the room. Of course, it doesn’t tell us the whole story, but it does give us a sort of representation of the story. It’s not so much a guide that will hold our hand through the mountainous terrain of dysphagia management, but it does at least tell us where we are on the map of that terrain. But this approach is only as good as the map we have and only as good as our ability to read that map (or our willingness to read it at all). My advice is to start small. Do some research on one patient each week to determine the base rate. Then see how that information can fit into your plan of care and how it may impact your assessment and management of the patient’s risk. It may not impact it at all. Or it may change your mindset completely. In any case, at the very least, it’s worth taking a look.
Liked this? Why not share it?
Leave a comment. I feed on feedback.
