Editor’s Spotlight/Take 5: How Is the Probability of Reporting Various Levels of Pain 12 Months After Noncatastrophic Injuries Associated with the Level of Peritraumatic Distress?

As someone whose practice consists principally of elective orthopaedic surgery, I’ve become absolutely convinced that the most important thing people like me get wrong is when we offer discretionary procedures to patients who have, along with their musculoskeletal conditions, poorly controlled anxiety or depression. Patients with untreated psychological distress are at greater risk of developing persistent pain, clinical dissatisfaction, or a serious complication after arthroplasty than are patients who smoke or those who have morbid obesity [9, 12, 13]—two conditions that surgeons screen for aggressively [10]. The association between psychological distress before surgery with persistent pain and clinical dissatisfaction afterward has likewise been observed in procedures performed by spine [15], shoulder [16], and sports specialists [3, 5].

It shouldn’t surprise us that the human condition trumps conditions of the musculoskeletal system, and that these findings probably apply to most or all orthopaedic surgery. Yet few orthopaedic surgeons I know screen for depression, anxiety, and related conditions. Many have told me they feel it’s beyond their expertise, or that it just takes too long to use the available screening tools.

Wouldn’t it be nice if there were an easy-to-use questionnaire that could identify which specific patients were at risk for persistent pain after the kinds of orthopaedic treatment that really ought to go well?

In this month’s Clinical Orthopaedics and Related Research®, a group from Western University in London, ON, Canada, share just such a tool. The Traumatic Injuries Distress Scale (TIDS) takes less than 3 minutes to complete, and in this study [7], it reliably anticipated which patients who presented with noncatastrophic orthopaedic injuries—defined as injuries not treated surgically or with hospitalization, events from which we expect patients to recovery fully and easily—were more likely still to have pain a year later.

The TIDS (Fig. 1) focuses on patients’ self-perceived relationships to pain, as well as issues of affect, intrusive thoughts, and hyperarousal. Importantly, you don’t need any special expertise to be able to use it. Patients self-administer the simple, short questionnaire, and 3 minutes later, surgeons have a robust measure of risk; no psychology consult or on-site expert needed. In this study, patients with minor injuries who scored less than 5 of 24 on the scale had only a 1 in 10 chance of still having pain 12 months after injury, while the risk increased to 30% among those who scored 13 or more.

Fig. 1.

TIDS questionnaire. This survey instrument may be reproduced and used for clinical purposes as long as it is not used for profit or modified without consent of the copyright holder. Email dwalton5@uwo.ca for inquiries. ©David M. Walton, Western University.

Obviously, investigators can and should replicate these findings in patients undergoing other procedures, including major surgery, but I have a strong suspicion the findings will persist across a wide range of conditions and operations. Human knees, shoulders, and spines split the rent with human souls, and if those souls are distressed, it probably matters little what procedures we perform on the bones.

As a worthy aside, the analytic approach (quantile regression) used by this research team, which was led by David M. Walton PhD (Fig. 2), is one I’ve not seen or used before, and it strikes me that this approach can help us answer a host of pressing problems in clinical research. But as with so many tools, it helps to know something about its strengths and soft spots.

Fig. 2.

David M. Walton PhD

For all those reasons and others, I hope you’ll join me as I go behind the discovery in the Take 5 interview that follows with Dr. Walton, senior author of “How Is the Probability of Reporting Various Levels of Pain 12 Months After Noncatastrophic Injuries Associated with the Level of Peritraumatic Distress?” [7].

I should add that this paper was published as part of the selected proceedings of the first meeting of the International Musculoskeletal Mental and Social Health (I-MESH) Consortium [11]; it’s accompanied by a number of other papers that I think are just as fun to read and informative. I hope you’ll enjoy them as much as I did. Because it was so successful, CORR® anticipates covering I-MESH-2 around this time next year.

And as always, I hope you’ll send your thoughts and questions about this paper to CORR in the form of a letter to the editor, which you can email to eic@clinorthop.org.

Take 5 Interview with David M. Walton PhD, senior author of “How Is the Probability of Reporting Various Levels of Pain 12 Months After Noncatastrophic Injuries Associated with the Level of Peritraumatic Distress?”

Seth S. Leopold MD: Congratulations on this important and thought-provoking paper. This is one of those studies that sent my mind racing in many directions at once because it offers so many ways to change and improve patient care. How have providers at your participating study centers started to incorporate your discoveries vis-à-vis the TIDS tool into their practices?

David M. Walton PhD: That’s a slightly more difficult question to answer than you might think. In the interest of equity to access, we made the decision some years ago to make the tool freely available to anyone interested—private, public, industry, or otherwise. While this is great for ensuring that useful work makes it into the clinical arena where it can be implemented to help patients, it has the disadvantage of meaning that we tend to lose track of just where it’s going and who is using it. I know, for example, that one of our team members, Dr. Modarresi, has recently completed a cross-cultural translation of the tool into a Persian version published online in the Journal of Advanced Medical Sciences and Applied Technologies [8], and I’ve just submitted a manuscript for review in which I worked with a clinician in Chile to create a Spanish version. The TIDS tool has also been programmed into the Focus on Therapeutic Outcomes (FOTO) online outcome measures platform [4], which is one of the largest platforms used by physical therapists in the United States. We’ve heard from clinicians in Canada, the United States, Ireland, Australia, and other regions who’ve indicated that they’re interested in using the TIDS. I should note that this is not the only such prognostic tool available, and I encourage your readers to seek out others such as the Keele STarT Back Tool [14] or the Orebro Musculoskeletal Pain Questionnaire [6] to determine the best tool for their context. I believe that prognostic screening tools are becoming increasingly used and that the best ones are those that provide clinicians with not only the magnitude of risk that a patient may not recover quickly, but perhaps even more importantly, the reasons for that risk, especially if those are potentially modifiable. That was the vision I had when initially developing the TIDS.

Dr. Leopold: I recently heard a podcast in which David Ring MD, PhD was interviewed. Dr. Ring is a Deputy Editor at CORR, as well as a co–guest editor of the CORR proceedings in which the current paper appears, and so I consider his insights germane on this topic. In that interview, Dr. Ring said that orthopaedic surgeons have an ethical duty to get the diagnosis of all the mental, social, and physical health opportunities correct; to correctly prioritize them; and to address each of them in parallel rather than in series [2]. From where I sit, the fact that so much elective surgery is performed on patients with untreated or inelegantly managed anxiety and depression is evidence that we’re not living up to this ethical imperative in practice. How might discoveries like yours help surgeons—who are not experts in the diagnosis or treatment of conditions that cause psychological distress—to do better?

Dr. Walton: There are a few different ways to consider this. First, it bears mentioning that not everyone with a mental health disorder will develop chronic pain, and not everyone with chronic pain has a mental health disorder. This is important because many of the current models of chronic pain development have centered on the mental health component, and as a result, some people have misinterpreted those models to blame patients who develop chronic pain for their own condition, as though they were somehow “thinking wrong”. This is partly where I hope the TIDS can nuance that conversation slightly. The subscales in particular, which were not a main focus of this paper, are intended to help clinicians better identify priority areas for intervention in those patients deemed higher risk of chronic pain. For context, those subscales are: Uncontrolled Pain (the best predictor of persistent pain at 12 months), Negative Affect (the best predictor of depression scores at 12 months), and Intrusion/Hyperarousal (the best predictor of PTSD-like symptoms at 12 months). Our hope is that even in the absence of a frank psychiatric diagnosis, clinicians can use a scale like the TIDS to facilitate early intervention; if the patient indicates that since the trauma or injury the patient feels less motivated to get up and start a new day or feels detached from the rest of the world, this should ideally cue a clinician to consider a consult to a mental health professional early so those feelings can be addressed before they become resistant to treatment. The other part of this answer is to remind all clinicians that the most recent estimates I’ve seen indicate that some 20% to 30% of the North American adult population lives with a mental health disorder [1], and at minimum, these should be acknowledged for their impact on the patient’s sense of health and wellness. Even better, we encourage clinicians to adopt more trauma-informed ways of practice when engaging with their patients to manage the experiences of distress from trauma, whether that be an injury or a surgical procedure.

Dr. Leopold: It’s probably not reasonable for a spine surgeon, a shoulder surgeon, or a knee surgeon to use the actual numeric thresholds in your paper (the specific TIDS scores that you found were associated with increased risk of persistent pain), since those thresholds were derived from patients with relatively minor injuries and who did not have surgery. At the same time, it seems very reasonable for them to begin to get some experience using the tool in their practices. How might they do this in a way that is practical and informative? Related to this, in the absence of specific data in those yet-unstudied clinical settings, is there other research to suggest what TIDS score might be “high enough” to trigger at least a conversation about delaying elective surgery and pursuing treatment for emotional distress in advance of orthopaedic surgical interventions?

Dr. Walton: I’ll be cautious in my reply as I wish to avoid editorializing here and would prefer to adhere to the data we have available. So, while I do believe there is value in administering the TIDS in the acute (within a few days to a few weeks) postoperative stage to identify those subscale areas that require greater attention as described above, as of right now, I am compelled to state that I cannot prove that position empirically, and that this represents a priority area of further investigation. However, for those who do wish to explore the relative value of different cut-off scores, I encourage readers to consider those reported in this paper, or seek out another recent paper we’ve published [17] for a description of other potentially useful cut-off scores.

I would add that a colleague of mine is currently working on a project in which the TIDS is being used to predict recovery trajectories following distal radius fracture with or without surgical correction, and I’d expect those data to be available sometime in 2022. I’ve also been in contact with some local spine surgeons interested in using it in their practice, but so far it seems the majority of use has been in rehabilitation settings. If readers are interested in implementing the tool within their settings and turning that into a formal research project, I encourage them to reach out to me directly: dwalton5@uwo.ca.

Dr. Leopold: What’s the best way to use the TIDS in the flow of a busy clinic, and is the approach to all patients with high TIDS scores pretty similar—is it always a psychiatry consult and/or phone call back to the patient’s primary care physician—or does it matter which questions on the TIDS resulted in the high scores? How do you parse out TIDS scores that you find worrisome? I pray that it’s easy, as some surgeons seem to be looking for a way not to make this our business, and I think we need to.

Dr. Walton: As with any such tool, the more you use it the quicker and easier it gets to interpret. However, I will say that our current analysis using quantile regression indicates that while full interpretation is somewhat more challenging than a simple sum, for the purposes of quick clinical use, a simple sum usually will do. Generally speaking, my recommendation is to score the items, and a total score of 11 or greater should spark a slightly deeper look. From there, I encourage clinicians to consider the subscale scores, which will answer part of the “why, and what should I do about it?” question. Fortunately, I’ve created a very simple Google Sheet for this purpose that anyone can access from here as long as they only enter scores in the gray box: https://docs.google.com/spreadsheets/d/1FG7S_hX1d1s59WcEbz9-50Isph7DGXJRGgw7qVceWEI/edit?usp=sharing. The question of what to do about it can be dictated by those subscales; if the patient’s distress seems to be driven largely by a sense that he or she is unable to control their pain, then looking at the strategies the patient is using now and working on a personalized pain management plan is a logical first step. If the patient describes considerable negative mood/affect, then a psych referral may be of value, though my suggestion would instead be to explore that with another simple screening tool, such as the Patient Health Questionnaire-9 (PHQ-9) as a depression screen, which as the name implies is only 9 items long. If that tool also indicates a major depressive disorder may be present, then I would think a clinician can partner with the patient to identify an appropriate path forward that may or may not include formal psychological counseling or psychiatric treatment. If Intrusion/Hyperarousal seems to be the dominant driver, then again, I would encourage the clinician to explore that further; perhaps administer something like the PTSD Diagnosis Scale or another such tool, which can be administered by a nurse or other support personnel. Importantly, however, I encourage surgeons to educate themselves on mental health issues and how to respond appropriately. They’re likely quite right that their training has not been adequate to prepare them to respond to disclosures of, for example, thoughts of self-harm or death. But recalling that somewhere between 1 in 5 and 1 in 3 patients they see daily will live with a mental health disorder, there are fewer and fewer reasons for them not to take the initiative to prepare themselves.

Dr. Leopold: Let’s conclude with your analytic approach, quantile regression, which I’ll admit was unfamiliar to me, and I’ve been in this chair for nearly 10 years. Why is it so well hidden, what do you see as its advantages and liabilities, and where is the low-hanging fruit; where will readers see it used next?

Dr. Walton: I’ll fully admit that this analysis was driven by my former PhD student Dr. Shirin Modarresi, and I too needed some time to fully understand it. We had already shown in prior work that the TIDS had prognostic value when predicting recovery trajectories after MSK trauma [17, 18]. Those prior analyses, however, assumed a fairly simple linear association, so Dr. Modarresi convinced me that quantile regression was a worthwhile secondary analysis. I’m glad she did, because I quite like what it has found.

Interestingly, the technique has been around since the 1970s, but only recently has found increasing use in health research, having been introduced into common statistical packages like SPSS in 2019. Importantly for my answer to this question, I believe there is both value but also potential danger in labeling a patient as “at risk,” so the more accurate we can be in how we interpret such scales the better. It bears mentioning, however, that all such results are probabilistic—there are many variables, both internal and external to the patient, that will affect their recovery beyond their scores on a self-report scale. Quantile regression allows the relationship between variables to vary across the entire population rather than relying on the mean of the data that ordinary least squares (OLS) approaches require. In practice, this means we are less constrained by interpreting research results in relation to only the “average” patient, whoever those people are. I believe that the common assumption of linear associations between different health indicators in humans is problematic, so the quantile regression approach has allowed us to explore these associations with much greater granularity.

Using our paper as an illustrative example [7], we found that if people report a low TIDS score within the first few weeks of an injury, and if they subsequently develop or report a chronic pain problem 12 months later, then the genesis of that chronic pain was probably not related to baseline distress; something else must have been at play. On the other hand, those who reported a high baseline TIDS score and reported persistent pain 12 months later showed a much stronger relationship, suggesting distress was a strong driving factor. If we were to put this in the context of orthopedic surgery, we know that joint arthroplasty outcomes tend to be good on average, but there remains a small subset of the population who experience problematic outcomes. This presents a problem in normal OLS-based linear regression due to the non-normal distribution of outcomes. Quantile regression relaxes these assumptions by modeling the entire conditional distribution of outcomes and predictors, offering protection against spurious findings due to non-normal data. As we continue to further explore these associations across different populations, a logical extension of this work will be to evaluate the effectiveness of addressing post-trauma distress in those with higher TIDS scores, while in those with lower distress, clinicians can probably take a more hands-off approach.