Evaluate Research Relevance

By Ravensara S. Travillian
[Somatic Research]

In our scenario, an older male client, who comes to you regularly for massage, was living with arthritis pain and asked you for recommendations—specifically, whether massage could be useful in supporting his health goal of minimizing pain medication.

You considered that, since massage was effective for musculoskeletal pain, it may be effective for osteoarthritis (OA) as well. We found an article about the effect of massage on reducing arthritis pain: “Massage Therapy for Osteoarthritis of the Knee: A Randomized Controlled Trial,” by A.I. Perlman et al.1 Finally, we left off with questions for evaluating the article’s relevance. Now we can turn to the questions and compare notes.

While I’ve given my answers here, many questions leave room for interpretation, based on different experiences in practice; consider factors that could account for these differences. I’d love to hear about your answers. Send your comments to researching.massage@gmail.com.

Extract the Basics

We asked a few questions about the article’s basic elements. Why did the researchers do this study (from the Introduction)? What is their research question? Does their context mean anything to my practice?

Perlman et al. conducted this study in order to “evaluate … the effectiveness of massage therapy for OA.” Their assumption was that massage would provide significant and detectable improvement, much in the same way as in their assertion that “[m]assage therapy has been evaluated and found to be effective for various painful musculoskeletal conditions.” They determined that a sample size of 66 subjects provided enough statistical power to detect a 20-point difference in pain scale scores between intervention and control groups at eight weeks; they recruited 68 participants. Their context is very close to our scenario, and it sounds like the researchers have taken care to create a study large enough to generate reliable results that may generalize well. For these reasons, this study is likely to be very relevant to our practice.

Patient, Intervention, Comparison, Outcome

Relevance of research is driven by how well the patient, intervention, comparison, and outcome (PICO) of the study relates to our own PICO question. We asked: how does their PICO compare to ours (from the Introduction)? The comparison is illustrated in Image 1, page 129.

Study, Findings, Results

How did the researchers study their design question?

Here, they used a randomized controlled trial (RCT) design, considered the gold standard of research methodology because of its ability to minimize bias in distinguishing among treatments. There were two groups: the control group and the massage group. In addition, the control group crossed over to receive massage treatment after the first part of the study was completed.

Although control groups sometimes receive absolutely no treatment, that was not true here. It is unethical to make subjects suffer by participating in a study, and since there is a standard of care for OA patients (pain medication, exercise, hot and cold therapy), massage is compared against that standard of care rather than against no treatment at all. It represents a common standard and a humane ethical compromise.

The difference between the standard care group and the massage group reflects another methodological issue. If the massage group showed an improvement over the standard care group, we would assume massage is what made the difference. But—as the fundamental question of science insists—how would we know if we were wrong? Perhaps simply having a caring human around for 20 minutes, whether providing massage or not, is enough to encourage healing. To sort this out, researchers often include a third group who get a visit—but no massage—from the massage therapist. If the researchers find a difference in outcomes between the individual attention group and the massage group, we would assume the difference was due to the massage itself, not just the presence of another caring person.

Here, the researchers’ description of how they carried out their study sounds reasonably good. The sufficiently large research population combined with the three-group RCT design makes this seem like a well-designed, robust, solid study.

What did the researchers find out?

The similarity of the numbers in Perlman’s Table 1 indicates the groups were similar to each other in age, socioeconomic status, and other demographic variables—meaning any differences in outcomes are unlikely to have been caused by personal differences among the groups. As Perlman’s Table 2 shows, the groups experienced similar levels of symptoms before the massage intervention, so each treatment presumably had about the same therapeutic work to do—meaning any differences in outcomes are unlikely to have been caused by different “starting points” among the groups. Table 3 displays those differences in outcomes, demonstrating that the massage group outperformed the other groups in a statistically significant manner (and at 16 weeks, demonstrating that the improvements lasted for a relatively long time after the interventions stopped), except for the measurement of range of motion in degrees.

What do the results mean?

In the last article, I asked: what do the researchers say their results mean (from the Discussion, which the researchers call “Comment” in this article)? What do you think of their interpretation? The researchers judge their results to mean that “massage therapy using the Swedish technique is safe and effective for reducing pain and improving function in patients with symptomatic OA of the knee.” Their effect sizes are large, indicating a strong treatment effect, and they provide a technical reason why they think their results may be conservative, actually underestimating the effects of massage. They place their results in the context of studies on massage for musculoskeletal conditions, asserting consistency with results demonstrated elsewhere.

As the researchers mention in their Introduction and reiterate in their Discussion, they consider their study to be extremely important. They cite statistics on the incidence and prevalence of OA, and the associated decline of quality of life, serious side effects (“cardiovascular, gastrointestinal, renal, and hepatotic toxic effects” associated with medications used to treat OA pain), and economic and other costs which result. The size of the study, and the confirmation of large and lasting effects of massage on OA pain and loss of function are factors that Perlman et al. consider very important in making their case. All these aspects made the study seem solid.

Apply the Knowledge

In our scenario, would this article have provided any useful information for you to give your client? If so, what?

These questions may lead to very individualized responses. I approve of the way Perlman et al. dealt with their massage treatment protocol, and how they balanced the need for standardization with the inherently individualized nature of massage. By providing a basic protocol, and checking to ensure that protocol was adhered to, yet permitting individual therapists to make judgments in a session about which strokes to use, and in which order, I consider this article to be more practical and useful than those that do not provide information about their treatment. Many research articles describe their treatment only in vague, high-level terms; I can imagine an experienced practitioner would consider that description sufficient, while a newer practitioner might appreciate more specific procedural information.

A general principle of scientific and medical research articles is that the author should give the reader information to replicate the work. A good article will, or it will point to another source. By that guideline, did Perlman et al. give you enough information to replicate the work?

Limitations and Strengths
Weak Points

In the last article, I posed questions about possible weak points in the study. What problems or gaps did Perlman’s team find with previous massage research studies (from the Introduction)? What limitations in their own study did they address (in the Discussion)? In your assessment, are any of these problems severe enough to invalidate their study?

They lost a “noteworthy” number of their study participants for follow up, meaning they couldn’t trace the long-lasting results with those participants. They point out that this limitation is an artifact of the real-world clinical experience of working with a population of this age, who experiences limitation in mobility and for whom the logistics of a study may pose a hardship.

Another methodological issue was the increased possibility of a Hawthorne effect (positive outcome bias on the part of the study participants due to attention from the researchers) because of their efforts to address the problem of “placebo massage,” another recurring basic methodological problem in massage research. While they believe that the intragroup and intergroup differences in outcomes (and the persistence of those positive effects) are an argument against that positive bias, it is worth attention.

Although Swedish massage is a common treatment technique, the parameters for choosing the type, number, and length of massages was a factor in designing this study. The researchers state that their approach is not necessarily the most effective, and this issue needs to be addressed further.

Study participants were overwhelmingly white women from a limited geographical range; this may limit how well the results can be generalized to more diverse groups with OA. However, these researchers note that the study is based on the population in which knee pain is the most prevalent. For a chronic condition, 16 weeks comprise a very short slice of time and longer-term studies would be appropriate. Additionally, they cannot reliably assess any change in usage of pain medications, because the study participants did not keep accurate medication diaries, although it is unlikely that massage would cause them to use more pain medication.

The Value of Structure

I posed the question: having read the article, do you think the Abstract is an accurate representation of it? I think the Abstract accurately represented the information in the article, and the structured format of the abstract made it easier to find particular information.

Structure also became my own answer to the question: what else
is notable?

I found the highly organized structure made Perlman’s article easier to read than many others. The headings they used inside each Introduction, Methods, Results, and Discussion (IMRaD) section gave an indication of what to expect, making it easier to understand the overall study.

Bring it Home

Now that we’ve both located research and evaluated its relevance to your work, hopefully you’ll find this a replicable process for your future needs. This article was carefully chosen to provide a good example for the exercise, but it’s not always that easy to find good evidence-based information for the questions that come up in practice. How much did our exercise reflect your real-life experience in finding research articles to inform your practice?

 Ravensara S. Travillian is a massage practitioner and biomedical informatician in Seattle, Washington. She has practiced massage at the former Refugee Clinic at Harborview Medical Center and in private practice. In addition to teaching research methods in massage since 1996, she is the author of an upcoming book on research literacy in massage. Contact her at researching.massage@gmail.com.


1. All quotations cited here are from Adam I. Perlman et al., “Massage Therapy for Osteoarthritis of the Knee: a Randomized Controlled Trial,” Archives of Internal Medicine 166, no. 22 (December 2006): 2,533–8.