Have you ever read an article that said one thing, just to read another that said the opposite? Although differences in study results are fairly common, they can be frustrating. How can we figure out the bottom line?
One approach to better understanding what the research says is to look for review articles. Review articles compile research articles into one big overview of a given topic. Review articles are frequently written by researchers and include the results of many different studies with a discussion about why the results may differ from one study to the next, as well as some commentary on the quality of the original research.  

In one popular article, Christopher A. Moyer, PhD, and his team reviewed literature to determine if there were consistent findings for a reduction in cortisol after massage therapy.1 Cortisol is known as the stress hormone, and many therapists and educators believe that massage treatment decreases cortisol levels, thereby also reducing anxiety, depression, and pain. However, the research literature had differing results; some studies demonstrated a decrease in cortisol with massage and other studies found no change. In the review article, Moyer found that while the studies demonstrated an improvement in anxiety, depression, and pain levels with massage, when the data were combined into one overarching analysis, there was no statistically significant change in cortisol levels. This review article helped clarify that perhaps there was another mechanism for feeling better after a massage other than a decrease in cortisol.

As you can see, review articles may help answer some questions—but they might raise new questions as well. The beauty of review articles is they give an overview of a topic area so you, the reader, can better understand the state of the evidence for that particular subject. However, not all reviews are the same; some are more rigorous than others.

Narrative reviews
One of the most popular types of review is the narrative review because it is the most basic. Narrative reviews are overviews of the literature without information on how the articles were chosen for inclusion. While narrative reviews can be very interesting, straightforward, and simple to read, they can also be very biased because there are no rules for how the reviewed articles are chosen. An author might choose his or her favorite articles that have a positive slant. Think back to high-school English class, when the teacher gave you an assignment to write about a topic using at least five research references. We all chose references that supported our point of view and ignored those that went against it. Narrative review articles have the same potential for bias.

An example of an interesting and helpful narrative review looks at the use of abdominal massage for chronic constipation. In 2011, massage therapist Marybetts Sinclair built upon a previous review of this topic by adding several more recent articles. The results demonstrated that “abdominal massage can stimulate peristalsis, decrease colonic transit time, increase the frequency of bowel movements in constipated patients, and decrease the feelings of discomfort and pain that accompany it. There is also good evidence that massage can stimulate peristalsis in patients with post surgical ileus.”2 Because this article is a compilation of several original research studies, the review is considered stronger evidence than an individual study. Information such as this may be of interest to your clients and can be shared with local doctors and surgeons. Even though Sinclair’s article was a narrative review and the methods were somewhat weak, the compilation of research articles gives a more thorough understanding of possible results of abdominal massage for constipation.

Systematic reviews
The second type of review article is a systematic review. As the name suggests, there is a systematic process used in this form of review. First, the authors state how the articles were gathered, such as by searching in PubMed.gov or other research databases. The key search terms are defined for the reader, as well as any limits that might have been placed on the search, such as “English articles only.” A systematic review will also define what type of articles are included—only randomized clinical trials, for example—and what types are excluded. Inclusion and exclusion criteria help the reader understand what type of research articles are being reviewed. Additionally, the systematic review will typically assess the quality of the original articles. Quality scores are presented in the results so the reader is aware of the strength of study design used in each original study.

The benefit of a systematic review over a narrative review is that the systematic review is usually higher quality and less biased. More information is provided in the systematic review, such as how and where to search for similar articles and which articles are the strongest. However, systematic reviews are typically longer because the description of how the review was done is described thoroughly before the overall results are described.

In the massage therapy literature, there are many systematic reviews. One such article by Myeong Soo Lee, PhD, Jong-In Kim, OMD, PhD, and Edzard Ernst, MD, PhD, assessed massage therapy for children with autism spectrum disorder.3 Twenty databases were searched for articles using the search terms “(massage OR touch OR acupressure) AND (autistic OR autism OR Asperger’s syndrome OR pervasive developmental disorder).” All forms of research were allowed, including case reports and case series, even though these lower forms of study design are not typically included in systematic reviews. Of the 132 articles found, only six were included in the review and most were found to have significant biases in the study design. The conclusion for this review was that “limited evidence exists for the effectiveness of massage as a symptomatic treatment of autism. Because the risk of bias was high, firm conclusions cannot be drawn. Future, more rigorous randomized clinical trials seem to be warranted.”

The need for more research is a common statement in review articles. In a 2011 article by Jenny Sau-Lai Chan and Sonny Hing-Min Tse, articles were assessed on massage therapy for relaxation and reduction of challenging behaviors in persons with intellectual disabilities.4 Sixty-four articles were found, but only seven were included in the review. Even though the majority of results were positive, there were many methodological problems with the study designs. “Case study designs, large amounts of qualitative data, and small sample sizes meant that the therapeutic effect of massage therapy could not be substantiated. Hence, future studies with randomized clinical trials or of experimental design are recommended,” Chan and Tse concluded.

Besides the need for a higher quantity of research, review articles frequently also call for higher quality. Wendy Moyle and her team wrote a review on the effects of massage on agitated behaviors in older people with dementia.5 The search revealed 13 studies that met the inclusion criteria, but only one study had adequate methodological quality to be included in the review. “This prospective study found that massage significantly reduced levels of agitation in 52 cognitively impaired residents in two long-term care facilities,” Moyle wrote. A stronger conclusion than this could be made if there were more articles and if they were higher quality study designs.

Meta-analyses
The final form of review articles is called the meta-analysis. A meta-analysis is a systematic review that includes an analysis of the data from the individual articles, such as the 2011 Moyer article on cortisol levels after massage discussed earlier. A meta-analysis will describe the method of searching the literature, what articles were included and excluded, and the quality of the included articles, and will then pool the data from the chosen articles into one major statistical analysis. If we look at the pyramid of study design hierarchy, the meta-analysis would be at the very peak. These reviews are the strongest types of articles you can read.

A recent meta-analysis by L. Wang and others assessed massage for preterm infants.6 A total of 611 articles were retrieved, with 17 eligible for review. The researchers found that “massage intervention improved daily weight gain by 5.32 grams and reduced length of [hospital] stay by 4.41 days.” The data presented in this statement were a compilation of the data from the 17 articles, making the results stronger than any one article standing alone.

Bias within review articles
Many factors can affect the results of a review article. For example, if the original research reviewed has flaws such as small sample size, then the results of the review article cannot be trusted. Some people call this “garbage in, garbage out.” Also, if there is a lot of heterogeneity (differences) in the original articles, the results may be muddied. For example, an author may review research on massage therapy for headaches and include articles on tension headaches, migraines, and cluster headaches in the same review. In this case, the original articles are studying very different conditions and the results cannot be lumped together.

One of the biggest issues in review articles is publication bias. Think about what articles get published in scientific journals: interesting articles with significant findings. But what happens to articles that do not have interesting results—perhaps articles where there was no difference in outcomes between groups? Sometimes they never get published! So, when someone is trying to review the literature for articles on a given topic, they might only be looking at the portion of articles with positive results because those are the articles that got published. Had the articles with nonsignificant results also been published, the results of some review articles might be very different.

Conclusion
Practicing massage therapists are busy people and many do not have the time to read all research articles published on a given topic. Review articles are great ways to get caught up on the current information in a succinct manner. They offer a great deal of information in one place, and can provide input on which original research studies might be worth reading based on the quality score. Systematic reviews and meta-analyses are also well respected by other health-care professionals. Sharing these types of reviews may improve other health-care professionals’ understanding and impressions of the massage therapy field.

Notes
1.    Christopher A. Moyer et al., “Does Massage Therapy Reduce Cortisol? A Comprehensive Quantitative Review,” Journal of Bodywork and Movement Therapies 15, no. 1 (2011): 3–14.
2.    Marybetts Sinclair, “The Use of Abdominal Massage to Treat Chronic Constipation,” Journal of Bodywork and Movement Therapies 15, no. 4 (2011): 436–45.
3.    Myeong Soo Lee, Jong-In Kim, and Edzard Ernst, “Massage Therapy for Children with Autism Spectrum Disorders: A Systematic Review,” Journal of Clinical Psychiatry 72, no. 3 (2011): 406–11.
4.    Jenny Sau-Lai Chan and Sonny Hing-Min Tse, “Massage as Therapy for Persons with Intellectual Disabilities: A Review of the Literature,” Journal of Intellectual Disabilities 15, no. 1 (2011): 47–62.
5.    Wendy Moyle et al., “The Effect of Massage on Agitated Behaviours in Older People with Dementia: A Literature Review,” Journal of Clinical Nursing 22, no. 5–6 (2013): 601–10.
6.    L. Wang, J. L. He, and X. H. Zhang, “The Efficacy of Massage on Preterm Infants: A Meta-Analysis,” American Journal of Perinatology 30, no. 9 (2013): 731–8.

Jerrilyn Cambron, DC, PhD, MPH, LMT, is an educator at the National University of Health Sciences and president of the Massage Therapy Foundation. Contact her at jcambron@nuhs.edu.