Is my patient getting better?
Am I missing anything?
How can I tell?
There were many moments early in my career where these questions were quite perplexing to me. I knew there was a way to find the answers, I just didn’t know where to look.
Then, during residency, a very simple treatment model was introduced to me. And it changed everything.
Finally, I had a simple way to determine how my patient was doing so that I could make informed decisions and have informed discussions with them. Not only that but the more I used it, the faster I progressed clinically.
The treatment model I’m talking about is the Test-Retest Model. It’s essentially the scientific method applied to clinical practice, and it’s how experts practice.
In this article we’ll cover:
Some of the questions & problems that we run into when it comes to treatment
Ways that the test-retest model can help us answer these questions
What the test-retest model actually looks like in practice
Some barriers to implementation
Situations in which the model might require adjustment
There are some technical steps in this post, so I took the liberty of creating a handy infographic for you. Click the link to scroll down to it. Enjoy!
Some of the problems we run into regarding treatment
The #1 problem that we run into with treatment is efficacy (the ability to produce the intended effect). In question-form, this would be something like: how do we know that what we’re doing is effective?
Do we treat every impairment that we find in the hopes that one will stick and the patient will improve?
Do we simply ask the patient if they’re getting better or not and take their word for it?
Do we wait 30 days until it’s time for a reassessment to see if we’re making improvements?
Another very important question to ask ourselves is: how can we develop pattern recognition skills so that we can get there faster in the future? Meaning, how can we learn from each of our patients so that the quality of our practice improves?
Answering these questions
The simple answer to both of these questions is: data.
Put simply: we cannot know if what we’re doing is effective without collecting data before & after our intervention.
Without testing and retesting, we are merely guessing that our interventions are working as intended.
Similarly, we cannot learn from our experiences without feedback. In fact, we really can’t learn anything without feedback.
Imagine practicing shooting a basketball without any feedback (e.g. blindfolded with noise-canceling headphones). You could shoot all day long and you wouldn’t get any better because you have no feedback, nothing to tell you if you’re successful or unsuccessful. Add even one piece of feedback—hearing—and you can learn. It would be tough, but you could hear if the ball hit the rim, hit the backboard, or missed entirely. Add another piece of feedback—vision—and now you can really learn.
It’s the same thing in clinical practice. If we don’t assess the efficacy of what we’re doing, we could practice for many years without getting much better.
There’s a phrase in the clinical reasoning literature that captures this quite well that states, "Expertise is not an inevitable consequence of experience (1)."
The bottom line is that we need feedback to get better. By testing, treating, and retesting, we are engaged in the scientific process that is driven by data, by feedback, so that we can treat people more specifically, more effectively, and so that we can grow as clinicians.
Other important benefits of this model
Expanding beyond classifications
At this point, many of us use one or several classification models to classify and treat people, whether we know it or not.
There’s nothing wrong with this—there are many benefits of using classification models. For example, they can help us to chunk relevant data to make informed clinical decisions. One weakness, however, is the fact that they are inherently limited. They can never fully capture the person because people are always more complex than the model.
So we may be treating someone in a typical impairment- or classification-based way, but when something isn’t adding up or the person isn’t responding as expected, this may be a sign that something outside the model is contributing.
This comes up a lot with things like fear or other cognitive/affective contributors to a problem. You may start by treating their relevant impairments, but if they’re not responding as expected, you may need to dig a little deeper to find out what else they’re experiencing.
The important bit here is that we have to know how they’re responding, and we'd only know how they’re responding if we're testing & retesting.
This is also very important from a safety standpoint. We can all easily recite the fact that failure to improve with conservative care can point to something non-MSK related, but if we don’t have the data to know if they’re improving, or by how much, we can’t make an informed decision on the matter.
Navigating The Fear of Reliance
This can also be very helpful when using passive treatments like manual therapy. One of the most common concerns people have with using manual therapy is patient reliance. But if an intervention—any intervention—isn’t working, and you have the data to back that up, it makes for a fairly simple conversation with the person,
“We’ve tried this and it’s not working, let’s work together to find something that’ll help you to move more toward your goals.”
In all likelihood, the person’s desire to reach their goals will be greater than their desire to receive passive treatment. And if not, they may be better served to see another professional who specializes in those services. Another simple (but not necessarily easy) conversation to have. Thanks, test-retest model.
Helping people learn
As mentioned in the episode & post about Law III of the 4 laws of quality care, I believe that teaching people how to learn and reason is one of the highest services that we can provide them.
What I love about the test-retest model is that it exposes the simplicity of PT. Because in reality, all we’re doing is using the data that we have to make a decision—a guess, really—about what we think will help, trying it, then retesting to see if we were right. Simple.
This is something that patients can—and should—be doing on their own, based on the tools we give them.
Just by nature of using this process (and helping the person to understand it), we are giving the person the gift of reasoning.
The big picture is that to make informed decisions and have informed discussions, we need data. If we’re not engaging in the process of data collection, hypothesis generation, and validation, we’re shooting blindfolded with noise-canceling headphones.
What Does The Test-Retest Model Look Like in practice?
Think of it in an hourglass shape:
You start broad by assessing the person’s activity limitations
You get narrower by assessing their relevant impairments
You then zero in on an intervention and provide the intervention
Then broaden back up by re-assessing their impairments and their activity limitations
So it goes from broad to narrow to broad—like an hourglass.
Let’s dive into each of these a bit more specifically, using an example of a 50-year-old who has left-sided low back pain (LBP) when walking for more than a few minutes.
Step I: Assess Activity Limitations
The first thing we must do is assess the person’s activity limitations (the functional tasks that provoke their symptoms).
In order to do this, we must first determine what their activity limitations are. This is part of the normal history-taking process and should be followed up on during follow-up visits.
After we determine what their activity limitation is, we have them perform the task. As they're performing the task, we are assessing for:
Reproduction of symptoms. Some call this an asterisk sign, comparable sign, or concordant sign (reproducing someone’s familiar symptoms). I’m going to use the term functional asterisk
If you notice anything (and they are having symptoms), you can modify the movement to see if it changes the symptoms.
Let’s visit our example patient:
You determine the person's primary activity limitation is walking
You have them walk (great job so far!) and you notice a left lateral lean at midstance, limited left-arm swing, and at the two-minute mark they start having their familiar pain at midstance
You cue them to swing their arms evenly, and you check to see what happens.
Are they actually swinging their arms evenly?
Does the left lateral lean change?
Do their symptoms change?
*If their symptoms change, you can follow that thread as far as you want, that’s your call.
Step II: Assess Relevant Impairments
In this step, you break down the movement into its component parts and test each part individually.
So for this example patient, what are the movement requirements of midstance during gait?
From a muscle power standpoint, we need the hip/trunk muscles to be working
From a mobility standpoint, we need knee extension, neutral hip extension, neutral lumbar extension
From a coordination standpoint, we need a momentum transfer to occur (from the upper quarter, from the contralateral leg)
Now we test each of the movement requirements individually, again assessing for asymmetries & reproduction of symptoms. We’ll call the impairment(s) that reproduce their symptoms the impairment asterisk(s).
In standing, you screen single leg stance as a functional measure of hip/trunk power and coordination
Also in standing, you screen lumbar mobility in extension, side-bending, and rotation
In supine, you screen knee extension mobility
In prone, you screen hip extension mobility
Also in prone, you screen lumbar mobility and/or movement or pressure sensitivity through unilateral and/or central PAs
*As you screen, you measure anything that reproduces their symptoms or is different from the other side. Get that objective data!
Step III: Treat Relevant Impairments
So you assessed the person’s activity limitation (walking), broke down the movement, tested the components individually, and now it’s time to treat!
What you decide to treat depends on a few things:
If the person presents with high symptom irritability, the general approach is to help them rest. This can be accomplished through treating distant impairments to reduce load on the painful area during the painful movement, movement modification to reduce load on the painful area, activity modification (pacing, taking breaks, etc.), and more.
On the other hand, if the person presents with low symptom irritability, the general approach is to help them reload. This can be accomplished through treating more locally in the painful area in order to load into the painful/limited movement through exercise, movement modification, manual therapy, and more.
2. Severity of impairments
For example, a marked knee extension mobility deficit will take precedence over a mild hip extension mobility deficit.
Whatever you decide to treat, perform the intervention dosed according to your irritability hypothesis (as discussed in the post on Law IV of the Four Laws of Quality Care). If you’re at all concerned about making symptoms worse, it's better to under-dose than overdose (you can always do more).
Step IV: RETEST!
Ah, yes! Finally! This is where the magic—and the learning—happens, on the patient’s end as well as on ours.
Nine times out of ten, I start first by retesting the functional asterisk. This is for a variety of reasons including convenience, expectancy, and other contextual factors—but that’s just me.
When you retest the functional asterisk, have the person perform the SAME functional task(s) under the SAME conditions as before. If they had shoes on and you took them off, put them back on. If they did a forward bend with their feet together, make sure their feet are together again. That’s just good science—apples to apples.
During the retest, you will assess for any changes in symptoms and your objective measurements.
I’m going to give you a little algorithm here as a guide, which is also outlined in the infographic below.
If the person’s symptoms with the activity limitation are unchanged, retest the impairment asterisk(s):
If the impairment asterisk is no different, repeat the treatment with more vigor (you may have under-dosed)
If the impairment asterisk is improved, treat a different relevant impairment (you changed the impairment—well done—but may have chosen the wrong impairment)
If the person’s symptoms with the activity limitation are improved, repeat the treatment with more vigor (it worked the first time, so do more!) or move on to exercise that addresses the same impairment or movement. Great work!
If the person’s symptoms with the activity limitation are worse, repeat the treatment with less vigor (you may be on the right track, but just overdosed)
The big idea here is that if we don’t change the person’s functional asterisk, we need to keep digging to see what changes were made on the impairment level so that we can make the right next decision.
Don’t just give up if you don’t change anything on the first try. This should really be called the test-retest-retest-retest…. model.
Barriers To Implementation
I’ve worked with a lot of people on implementing this model, and there is one common thread that comes up in nearly everyone who is new to this treatment model: vulnerability.
The fact of the matter is that treating in this way is vulnerable. This is because when you come back and retest, you might be wrong. And if (when) you are wrong, both you and the patient will know it. And that can be uncomfortable. This discomfort is enough to make people retreat into old habits.
But don’t let this dissuade you. If you can tolerate the discomfort of this, you WILL get better. If you consistently treat using this model (assuming you’re testing well, objectively measuring, & not asking leading questions), you will effectively become your own mentor.
And it’s also in this vulnerability that deep trust is created. Engaging in this model is saying to the person, "I’m more interested in helping you than I am in coming off a certain way." You’re essentially prioritizing the person over your ego.
I’ve had many experiences with people who I couldn’t help give me a heartfelt, “Thank you.”
“For what?” I wondered. “I wasn’t able to help.”
But now I get it; they were thanking me for listening, hearing them, and trying my best—that’s all we can do.
And on the flip side, when you DO make a change upon retesting, it helps build hope, buy-in, rapport, and the look on the person’s face is a nice bonus!
The other common barrier is that we sometimes don’t believe that things can change within a single session. This is common and makes sense if you’re treating based on a pathoanatomical diagnosis or you believe that structural changes need to take place before improvement can occur. If that's the case, why would you retest?
My advice would simply be to try and see for yourself. And if you’re interested in exploring this idea further, give a read to the Steiger article below (2), and good on you for keeping an open mind.
When this model might not be useful
If it won’t change what you do
If you’re showing somebody simple ways to get moving, this might not apply. If you’re simply training someone on an assistive device, this wouldn’t apply.
But I would argue that there are be very few times when what we do isn’t based on some sort of testing. And if we’re testing, it only makes sense to retest following the intervention.
Those with very high or very low symptom irritability
Very high symptom irritability. Going through this entire process with someone with very high symptom irritability doesn’t make much sense. They don’t need to suffer through that and it would probably make them worse. In these cases, we'd likely be assessing/reassessing their resting symptoms, trying to help them find positions of comfort, and finding ways of helping them move that don’t make the symptoms significantly worse.
Very low symptom irritability or someone whose symptoms you cannot reproduce in the clinic. For these folks, first consider a non-MSK pathology. It should always raise our brows when we can't reproduce someone's symptoms with testing. Second, never underestimate the power of overpressure :)
But assuming you cannot reproduce their symptoms no matter how hard you push and mash, the adapted version of this model is simply spreading it out over two sessions.
You still assess their movement and relevant impairments, but you will likely have to focus on what you find through observation and measurement vs. reproduction of symptoms.
Then, you provide a specific intervention (or a few interventions aimed at the same impairment), have them go do their activity that brings on the symptoms, and report back at the next visit. You can still get the data, it just takes a little longer.
The answer to the question of how do we know that what we’re doing is effective? is by collecting data — by testing, treating, and retesting
The test-retest model can be visualized as hourglass-shaped,
Starting broad by assessing the person’s activity limitations
Narrowing by assessing relevant impairments
Zeroing in on a treatment
Broadening back up by reassessing the asterisk signs that were collected
Using this model can feel vulnerable in the beginning, but stick with it and the quality of your practice will skyrocket
The model should be adapted to the person,
For people with very high or very low symptom irritability, it won’t look the same as it would for those with more moderate levels of irritability
Moulton, C-A., Regehr, G., Mylopoulos, M., MacRae, H. (2007). Slowing down when you should: A new model of expert judgement. Academic Medicine, 82(10).
Steiger, F., Wirth, B., de Bruin, E. D., & Mannion, A. F. (2012). Is a positive clinical outcome after exercise therapy for chronic non-specific low back pain contingent upon a corresponding improvement in the targeted aspect (s) of performance? A systematic review. European Spine Journal, 21(4), 575-598.