Editor’s Spotlight/Take 5: Persistent Tennis Elbow Symptoms Have Little Prognostic Value: A Systematic Review and Meta-analysis

“The patient failed conservative management, and so surgery was required.”

Heard that one lately? I’m guessing so. The idea that “failure” of nonsurgical treatments justifies offering patients an elective procedure (assuming we can think of one to offer, and we usually can), is as close to a bedrock principle of orthopaedic surgeons’ practices as I can think of.

It shouldn’t be. This thought pattern, widely shared among surgeons, presupposes all sorts of fallacies, and uses linguistic tricks to tilt the playing field towards doing more surgery [8]. This thought pattern also assumes we know more than we usually do about the natural history of the conditions we treat, the treatments we offer, and the procedures we perform. It deemphasizes placebo effects [3] as well as opportunities for patients to adapt to their conditions or to have equally important conditions (such as untreated psychological distress) addressed in a clinically appropriate sequence [1]. It confuses discretionary interventions with life-or limb-saving ones (shoulder arthroscopy is seldom really “required”; neither is knee replacement or bunionectomy, but they’re often referred to that way in orthopaedic journals), it leaves patients responsible for “failing” something they couldn’t possibly control, and it leads to patients and surgeons expressing preferences for risky, invasive procedures that may be both ill-informed and poorly considered [9].

There are so many things wrong with our common approach on this point of practice that it’s hard to know where to begin.

To our rescue comes a Finnish research team led by Teemu Karjalainen MD, PhD. In this month’s Clinical Orthopaedics and Related Research®, Dr. Karjalainen’s study group hosts the barbecuing of one of our specialty’s sacredest cows [2].

Teemu Karjalainen MD, PhD

Cleverly, his group meta-analyzed the placebo (or no-treatment control) arms of randomized trials about tennis elbow, and discovered that some 90% of people with untreated tennis elbow achieved symptom resolution by 1 year. They also found that the probability of recovery remained fairly constant over that timespan regardless of prior symptom duration. This finding undermines the concept that surgery is somehow indicated if a patient’s symptoms persist for some arbitrary duration of time, as is so commonly suggested. These meta-analysts correctly suggest that their discovery supports hand holding over trigger pulling, since persistent symptoms are as likely to resolve later as they were right after symptoms first appeared. Because of this, any proposed surgery would have to be effective fairly promptly and associated with little risk, if it is to beat the natural history of this common (and commonly misunderstood) condition.

The Editor’s Spotlight/Take 5 section of CORR®—the section you’re reading now—is reserved for the study I believe has the broadest potential general appeal. Readers therefore may reasonably ask: Why shine the spotlight on a condition that most orthopaedic surgeons seldom or never treat?

Three reasons:

1. This study calls into question an approach in practice—the concept of “failing” nonsurgical treatment after some arbitrary duration of time—that surgeons use almost across the board in our specialty, and it should make us wonder whether this basis for decision-making that we use every day is as sound as we believe. My guess is that it isn’t, at least not for all the conditions in which we use it; this is a message all orthopaedic surgeons need to consider.

2. This study should raise our consciousness to the fact that a host of factors we seldom consider probably play a large role in patients’ wellness as they encounter a wide array of chronic orthopaedic conditions, as well as some acute ones. These factors include regression to the mean (most conditions we treat wax and wane, and, apparently, some—like tennis elbow—go away on their own), as well as the idea that patients come equipped with very capable internal pharmacies [1] and have a vast ability to accommodate to symptoms and adapt to changes to their bodies if we’re willing to listen and help them do so [3]. As importantly, the fact that symptoms persisted beyond a year in a minority of patients in Dr. Karjalainen’s study [2] may be less of a reflection of any specific musculoskeletal pathology and more a function of other treatable factors like psychological distress [7].

3. This study employed a methodological approach—interrogating the control groups of randomized controlled trials to divine a condition’s natural history—that holds very generalizable promise. It should be performed for as many other conditions as we can think of, in all orthopaedic subspecialties. We need to remember that some conditions, including some very-bothersome ones, get better on their own. As we study more conditions using this model [2], I believe we’ll find that we know less than we think we know about those conditions’ natural histories. And as we learn more, we’ll become a little less enthusiastic about recommending some of the procedures we now perform all the time, particularly those—like knee arthroscopy for degenerative meniscal tears or meniscal symptoms [5]—that may be working more as a placebo intervention than a mechanical “fix” to any particular problem, or that have effect sizes no different from those of a surgical placebo [10-13].

This is important stuff. Don’t miss the chance to go behind the discovery to learn more about it with Teemu Karjalainen MD, PhD, senior author of “Persistent Tennis Elbow Symptoms Have Little Prognostic Value: A Systematic Review and Meta-analysis” in the Take 5 interview that follows.

Take 5 Interview with Teemu Karjalainen MD, PhD, senior author of “Persistent Tennis Elbow Symptoms Have Little Prognostic Value: A Systematic Review and Meta-analysis”

Seth S. Leopold MD: Congratulations on this important study. I don’t believe I’ve ever done an operation for tennis elbow (not even during residency), but I am hard pressed to remember a study about which I was more excited to publish than this one. So many important messages here. Orthopaedic surgeons pride themselves on treatment of “objective” pathology; the problem is that so much of that pathology is misunderstood or misconstrued. The “black disk”; the “partial-thickness rotator cuff tear”; the “degenerative meniscus tear”; each of those lesions looks malign, but they may not be responsible for all the ills we project onto them. The genesis of some of our misunderstandings may be our excessive hang up on the biomedical model over its biopsychosocial alternative. If “tennis elbow” is a lesion, and some have suggested it is, why does it just go away? Or is something else going on here?

Teemu Karjalainen, MD, PhD: This was an interesting project that started accidentally. We were planning a trial but thought that perhaps there are tendinopathies that resolve and tendinopathies that do not. Because of that, we felt we needed first to identify those who do not recover to randomize this particular population to surgery versus nonsurgical treatment, since it doesn’t make sense to randomize—and so perhaps perform surgery on—those who improve spontaneously. When we started digging into this by looking at placebo arms in various trials of patients with tennis elbow, we found a trajectory that resembled a half-life curve suggesting that no matter who we choose, they are likely improve too quickly in nonsurgical treatment arm to demonstrate a clinically relevant effect for surgery.

I don’t think we know why many degenerative conditions tend to fluctuate and so often resolve. It is obvious that current biomedical models fail to explain the variation in symptoms, and therefore finding biomedical solutions will be difficult.

Dr. Leopold: Musculoskeletal specialists are often so biomedically focused that they can interpret improvement in supposed placebo cohorts such as sham injection as potentially having benefited from the needle or the saline or some other physical aspect of the simulation. Your analysis throws cold water on that concept; your discovery should cause us to pause, and think, “Let’s take a moment to appreciate the amazing health capacity within each human.” Because, in the end, that’s what the placebo effect is—it’s the natural human inner pharmacy, put to good use. What factors make people—surgeons or patients—prefer to think the solution was external and passive, rather than internal and related to a participatory, active patient-clinician relationship?

Dr. Karjalainen: It is mostly cultural. We see symptoms as biological or mechanistic problems that can be solved by treating the root cause because this is what we were taught. That is not a bad thing; modern medicine often works—but not always. What we tend to overlook is that our body and mind have great healing and adaptive potential, and consequently, we do attribute too much of the observed improvements to the interventions we perform. We think it’s either the therapeutic step of the procedure, or then it is some physical portion of the placebo intervention, or some other element of something that we performed. But I suspect a great deal of the improvement we observe in many musculoskeletal conditions is just the condition’s own natural course and regression to mean, and not a placebo effect at all in the sense that we understand placebo effects. In other words, I believe that often we would see the same improvements even if we didn’t do anything.

Dr. Leopold: Surgeons (and sometimes patients) sometimes seem to consider the sharing of information, reorientation of common misconceptions, compassion, and companionship as “no treatment.” Does treatment always need to take a physical form, and if not, what are some good examples of treatments that don’t take a physical form that surgeons should use more frequently in their practices?

Dr. Karjalainen: Surely assurance helps to cope with the problems although it is “no treatment” on the tendon/tissue level.

Most importantly, though, I think we should stop “trying” treatments that have questionable efficacy when people are frustrated with persistent symptoms. Vast amounts of resources are wasted on treatments that do not work, such as PRP-injections, because we are unable to resist our willingness to help. Instead, take a few minutes to explain to patients that their bodies have a lot of potential to repair themselves, and reassure patients they can come back if the worst-case scenario (no recovery) occurs. This, of course, applies only to situations where we don’t have effective treatments. Tennis elbow is a such a good example of this; so many trials, and still nothing that really has been shown to work.

Dr. Leopold: I’ve had so much fun thinking about the big picture here that I almost forgot to deal with your actual subject: Tennis elbow. Based on what you’ve learned, how do you treat it now?

Dr. Karjalainen: What I’ve been doing lately with people with typical tennis elbow symptoms (I find they often come after 3 months to 6 months of pain), is that I draw the curve we identified in the study on a piece of paper and explain the prognosis with no active treatment. In short summary, they have about a 50% chance of improvement during the next 3 months to 4 months, and this rate continues for up to 1 year.

I then explain that with surgery, the best estimate is that they can expect a similar prognosis, meaning that natural course is hard to beat. Unfortunately, good studies to answer the question of whether surgery would result in faster recovery do not exist.

Occasionally, a patient still will be interested in surgery. Most aren’t. I see maybe one patient a week, and I’ll admit here that I have operated on one this year—a patient who wasn’t able to work, went back to work after surgery, and was happy with the result. But I also admit that this probably would have happened at some point without surgery, too.

Dr. Leopold: The data in this meta-analysis don’t apply to other operations, but I think your study approach can be very broadly applied in ways that can transform how we practice elective orthopaedic surgery. Let’s end with where we should go next: Where do you see the overlap of the greatest need (we can define that as frequent use of discretionary surgery in the face of inadequate knowledge of the natural history of the conditions being treated) with sufficient available evidence (enough randomized trials with no-treatment control groups) to support using your method again?

Dr. Karjalainen: Surgeons (and physicians more broadly) are notoriously poor at abandoning established and entrenched clinical practices. It seems that we inappropriately place the burden of proof (that is, the burden to show that a particular procedure lacks efficacy) on skeptics, and our default position is to continue performing operations because our personal observations suggest that there are benefits. Our anecdotal experiences cause us to look for and find flaws in trials suggesting no benefits, whether those flaws are present or serious enough to disqualify the main findings of those trials [4]. But, as I mentioned, it is often not the intervention that causes the improvement. We need a paradigm change. We should require rigorous evidence to show surgery (or any treatment) works before we implement it.

Another key takeaway specific to tennis elbow is that since this population has such a good prognosis, it will be hard to find effective treatments. Specifically, it seems unlikely to me that surgery would be the answer since recovery from surgery often takes months and adverse events do occur. But it is possible. We may find a transient benefit in some subpopulation, but this has yet to be proven. Thus, we need a large rigorous trial comparing surgery with nonoperative treatment. There is one ongoing study [6] whose results I am really looking forward to reading.