Friday, May 3, 2013

Reflexology: Implausible Therapy for Pain?

Foot ReflexologyAs a form of complementary and alternative medicine (CAM), reflexology has gained popularity in the treatment of various conditions, including those relating to pain. Foot reflexology in particular, as a special type of massage therapy applying principles of Traditional Chinese Medicine, has generated much interest, and a newly reported study examined its effects on pain threshold and tolerance. Although outcomes were reportedly favorable, this and other research on foot reflexology does not appear to provide convincing evidence that it is an effective and durable treatment for any type of pain-related condition. Could this be an example of “Fairy Tale Science”?

Reflexologists propose that a system of reflex areas, or energy zones, reflect an image map of every part, gland, organ, and system of the body on the feet, and specialized foot massage techniques can produce physical changes in the body. An aligned hypothesis posits that meridians, which are essential landmarks for acupuncture throughout the body, converge at the ears, hands, and feet where they can be selectively stimulated by various techniques to produce desired effects that redress imbalances in “energy” that cause illness and pain.

Foot reflexology involves pressure applied to the feet using specific thumb, finger, and hand techniques without the use of oil or lotion. It allegedly dates back to ancient Egypt, but was introduced more recently into Western societies during the early 20th Century. The use of reflexology has increased considerably during the last decade and claims have been made for its use in treating such disorders as infertility, anxiety, depression, and pain.

To better explore the efficacy of foot reflexology for acute pain, researchers at the University of Portsmouth, United Kingdom, conducted a controlled experiment [Samuel and Ebenezer 2013]. They enrolled 15 healthy volunteers (11 females and 4 males, mean age 38) in a trial that induced an acute-pain stimulus and then assessed the effects of foot reflexology in moderating the pain.

  • Pain was induced by having subjects immerse their non-dominant hand in a container of ice-slurry while two indices of pain were assessed: threshold (the time taken for subjects to experience the first pain sensation), and tolerance (the time when subjects were unable to tolerate any further pain).

  • After baseline measurement of threshold and tolerance levels, each subject was assigned to either (A) an active treatment condition, consisting of a 45-minute foot reflexology session, or (B) a sham-control condition, consisting of a fake TENS (transcutaneous electrical nerve stimulation) device attached to the wrist and forearm above the non-dominant hand for 45 minutes, during which time no actual electric current was delivered.

  • Following the respective treatment periods, subjects immersed their non-dominant hand in the ice water at intervals of 30, 60, 90, and 120 minutes post-treatment, and data for threshold and tolerance levels were recorded.

  • All participants attended 2 experiment sessions 1 week apart, and a cross-over design was used so that each received 2 treatments of sham-TENS (control) and 2 treatments of reflexology in a random fashion. Data were evaluated using analysis of variance (ANOVA) for repeated measures.

Writing in the May 2013 edition of the journal Complementary Therapies in Clinical Practice, the researchers report that, compared with control data, reflexology significantly increased the overall acute pain threshold (F[1,14] = 4.60, p<0.05). However, a significant difference between groups was noted for hand submersion only at the 60 minute post-treatment interval (p<0.05), but not at 30, 90, or 120 minutes.

In terms of acute pain tolerance, the researchers similarly found that there was a significantly favorable overall effect of foot reflexology compared with the sham-TENS control (F[1,14] = 5.1095, p<0.05). There were significant differences between groups (p<0.05) at all but the 30-minute time interval.

Samuel and Ebenezer conclude that their findings demonstrate for the first time in a controlled study that foot reflexology produces acute-pain moderating effects, and that “reflexology may be useful on its own or as an adjunct to medication in the treatment of pain conditions in man.” However, there were numerous limitations and irregularities in this small study that merit closer scrutiny.

COMMENTARY: Ordinarily, we would not discuss a small laboratory experiment such as this, which the authors describe as an “exploratory” trial. However, this study offers some important lessons in principles of evidence-based medicine and it also sets the stage for a broader examination of reflexology as a CAM therapy for pain.

A first limitation of the study by Samuel and Ebenezer was that their trial involved healthy participants who were subjected to laboratory-induced pain, without the normal physiological and psychological attributes suffered by patients with either clinical acute or chronic pain. Second, it is commonly recognized by experts in evidence-based medicine that a research design like this — involving small numbers of participants, with repeated measures in a cross-over design — is strongly biased to favor a larger effect than warranted of the experimental intervention over the control condition.

Samuel and Ebenezer report strong overall effects favoring foot reflexology in their ANOVA calculations; however, in what they describe as “post-hoc tests” there was a significant difference between groups for pain threshold favoring reflexology only at the 60-minute interval. This difference apparently did not persist and was not significant at 90- and 120-minutes.

Additionally, the researchers report that significant differences between groups in pain tolerance favoring reflexology were evident at the 60-, 90-, and 120-minute intervals. They do not provide the raw data that went into these post hoc analyses, but they do show a graph depicting mean changes in pain tolerance for each group at the different time intervals, with error bars reflecting SEMs (standard error of the means; black vertical lines) — see figure.

Reflexology GraphAt first glance, the graphic portrayal seems to clearly suggest that the two groups are significantly different, starkly favoring reflexology for increasing pan tolerance times from baseline. However, a better visual indication of statistical significance would be facilitated if the error bars represented 95% Confidence Intervals (CIs) rather than SEMs.

Since, the 95% CI is roughly equivalent to the mean ± 2 x SEM, we drew lines (green and red vertical lines; each is 2x the respective SEM) that approximated the contiguous CI error bars for the 90-minute-interval data in the graph. As can be seen, the two CIs clearly overlap each other, suggesting that the difference in pain tolerance between groups at this time point actually is not statistically significant. The same sort of overlap, and lack of statistical significance, would be found at the other 2 time points (60- and 120-minutes) that the researchers claimed were significant. [These concepts of SEMs and CIs were discussed in a prior UPDATE here.]

It is impossible to know how the researchers arrived at statistical conclusions that are inconsistent with their own data depiction. However, this is an interesting example of how readers can assess for themselves questionable outcomes in research reports when very limited data are provided to substantiate results.

Beyond the data, there are other puzzling aspects of this small study. For example, the researchers describe the reflexology technique that was used as follows…

“A variety of movements are used in reflexology and these include holding, pressing, sliding, gliding, walking, stretching and rotations. The most often used technique is known as caterpillar walking in which the medial aspect of the thumb creeps slowly along in an intermittent on/off pressure. The technique used was that of the Bayly School of Reflexology and follows a basic treatment sequence of 45 min duration. A standard treatment method was adopted to ensure all reflex points were covered as pain is so subjective and none of the reflexology charts indicate specific reflex points for the hand or foot itself.”

It is interesting that the researchers concede that standard reflexology references do not denote specific reflex points serving the hand, which seems to detract from the scientific basis for their experiment and reflexology itself. Their procedure covered all reflex points, which does not support the specificity of reflexology principles. Along those lines, the researchers note elsewhere that foot massage by untrained persons is unsuitable as a sham-control condition because of its strong similarities to reflexology, which would confound results. This further questions the specificity of reflexology techniques, beyond what a soothing massage of the feet might offer in general.

Furthermore, Samuel and Ebenezer noted that they considered a 1-week period between sessions (washout period) was suitable because “there is no scientific evidence regarding the post-treatment effects of reflexology.” This seems to imply that, at best, the durability of pain moderating benefits of reflexology are quite limited, which might have been substantiated if the researchers had done assessments beyond the 2-hour recording period in their experiment.

So, even if the researchers’ overall statistical analysis is correct — that foot reflexology exerts a favorable influence on pain threshold and tolerance — it is unclear whether the procedure itself produced the effects and whether those benefits were more than fleeting. Still, this was only one small experiment of questionable quality, so it is important to also consider research evidence for reflexology from a broader perspective.

Systematic Reviews & Meta-Analyses of Foot Reflexology

There has, indeed, been considerable research on foot reflexology and these trials have been summarized in systematic reviews and meta-analyses. In a relatively recent study of this sort, a research team from several schools of nursing in Korea evaluated the effectiveness of foot reflexology on fatigue, sleep, and pain [Lee et al. 2011].

An extensive literature search by Lee and colleagues discovered 44 eligible studies, including 15 associated with fatigue, 18 with sleep, and 11 with pain. Writing in the Journal of the Korean Academy of Nursing, the researchers report that, although there was considerable heterogeneity across studies, there were large and significant pooled effects sizes (Cohen’s d [95% CI]) of 1.43 [1.50-1.61], 1.19 [1.03-1.34], and 1.35 [1.01-1.70] favoring foot reflexology as beneficial for fatigue, sleep, and pain, respectively.

While these results appear to strongly recommend reflexology as a CAM modality, it is important to also consider the quality of evidence that went into this otherwise excellently reported review and meta-analysis. Looking specifically at the studies included in the analysis of reflexology for pain, here are several observations…

  • The 11 studies of foot reflexology for pain had substantial heterogeneity (I² =59.19%), suggesting that there were important inconsistencies in quality, methodology, and available data across studies.

  • Only 3 of the studies had actually been published in (somewhat obscure) peer-reviewed journals; the others were master’s theses or doctoral dissertations.

  • Intervention periods among the various studies ranged from 1 day to 6 weeks, and the number of reflexology sessions ranged from 1 to 12.

  • All studies had intervention and control groups, but none of them qualified as randomized controlled trials (RCTs); most typically, randomization of subjects was absent, but other deficiencies also were evident.

  • Most important, group sizes were small, ranging from 11 to 22 subjects in 10 of the studies, and only about 30 subjects per group in the remaining and largest study.

These deficiencies are typical of those also found in the studies that examined fatigue and sleep. Of special interest, all of the studies in the review and meta-analysis were of relatively small size (≤30 subjects). A recent analysis in the British Medical Journal [Dechartres et al. 2013] — reporting on how trial sizes influence treatment effect estimates — found that meta-analyses of trials enrolling fewer than 50 subjects per group may overstate the true effect size by as much as 32% to 48%, as compared with larger trials.

Even if we reduce the large effect sizes favoring reflexology in the Lee et al. meta-analysis by 48%, the effects still would be at least moderate in size. Added to this, however, would be the many other limitations and concerns (listed above) that may have biased and skewed outcomes in the meta-analysis; together, these strongly reduce the level of confidence in the quality, strength, and veracity of this evidence supporting reflexology.

In broader-ranging systematic reviews of reflexology for a variety of conditions, and focusing only on bona fide RCTs, Edzard Ernst, MD — of Peninsula Medical School, University of Exeter, United Kingdom — and colleagues report disappointing outcomes for this CAM modality in 2 systematic reviews:

  1. In the first review, Ernst [2009] found only 18 RCTs meeting inclusion criteria, examining a range of conditions such as back pain, headache, irritable bowel syndrome, asthma, diabetes, cancer, multiple sclerosis, and others. Only 5 RCTs yielded positive results. Methodological quality was often evaluated as poor, and sample sizes were generally low. Most higher-quality trials did not generate positive findings.

  2. In a second, more recent and updated review, Ernst et al. [2011] located 23 RCTs meeting inclusion criteria and relating to a wide range of medical conditions (most of the studies overlapped with the first review). Methodological quality of the RCTs was often poor; overall, 61% of the studies (14/23) failed to show that reflexology was an effective therapy, while 8 RCTs suggested positive effects, and 1 was unclear as to the direction of results.

    Many studies were confounded by complex designs and questionable sham or placebo control conditions. For example, a study of foot reflexology for headache enrolled 32 patients assigned to either: (A) 12-to-30 sessions of reflexology + oral placebo, or (B) the control condition, consisting of arm massage + oral flunarizine (a calcium-channel blocker), during a period of 2-3 months. There were no differences found between groups on intensity and duration of headache.

    The largest RCT enrolled 243 participants in a study of chronic low back pain. Subjects were randomized in a cross-over design to 6 reflexology sessions, relaxation, or usual care. At 6-months followup there were no differences between groups in pain or functionality.

Ernst and colleagues conclude in both of their systematic reviews that the best available evidence “does not demonstrate convincingly that reflexology is an effective treatment for any medical condition.” They do allow, however, that the poor quality of many existing studies and paucity of high-quality studies (or their replication) prevent definitive judgments as to the value of reflexology.

Skepticism & “Fairy Tale Science”

Integrative CAMCAM modalities as components of integrative pain management are gaining greater traction, particularly in today’s climate of growing concerns about long-term pharmacotherapies in treating chronic pain conditions. Healthcare providers and their patients with pain are eagerly searching for alternatives to drugs, which also invites controversy since some skeptics claim that “integrative medicine” is just the latest buzzword for a collection of superstitions, myths, or pseudoscience that have gone by various names through the years. [This was previously discussed in an UPDATE here.]

So, where does reflexology fit in?

While few would dispute that a vigorous foot massage can be a pleasant and relaxing experience, possibly engendering some favorable albeit fleeting healthful effects, the evidence for recommending reflexology as a clinical pain management practice seems weak and generally unconvincing. Furthermore, one must look beyond empirical evidence from limited studies to consider the scientific plausibility of reflexology.

The historical roots and growing popularity of reflexology are not adequate arguments in favor of this modality from an evidence-based scientific perspective. And, the science skeptics have not spoken kindly of reflexology, asserting there is no anatomical or physiologic justification for any beneficial effects [eg, see posting by Mark A. Crislip, MD, at the Science Based Medicine blog here].

In their paper described above, Samuel and Ebenezer [2013] concede that the mechanisms involved in reflexology are undetermined. They speculate that alleged pain-mediating effects may be related to the release of endogenous neurotransmitters or modulators — like opioid peptides, serotonin, or noradrenaline — although, these have not been measured during laboratory or clinical trials. The meridian and energy-balancing theories applied to reflexology may have some appeal due to their affiliation with acupuncture; however, strong and enduring benefits above and beyond placebo effects in this regard have been challenged in thorough systematic reviews by Edzard Ernst and others [see discussion in UPDATE here].

In much of the research and writing on CAM therapies there is a plausibility problem, and the acceptance of certain modalities seems predicated more on faith or anecdotal experience than on valid and reliable scientific evidence. Increasingly, the scientific community is incorporating principles of Bayesian inference, proposed more than 200 years ago, for helping to assess the strength and validity of research evidence.

In essence, a Bayesian approach takes into account the prior probability, or scientific plausibility, of a hypothesis before assessing clinical trial outcomes and their validity. This prior probability is estimated from principles of natural science, the quality and strength of all previous research evidence (if any), biases that may have influenced investigators and affected the veracity of their work, and other factors. If the prior probability for some effect of an intervention or treatment is low, then far more evidence of good quality would be required to substantiate any claims of efficacy. In other words, the more extraordinary the claims, the more extraordinary the evidence that is required to validate them.

At its extreme, the promotion of therapies with unsubstantiated scientific plausibility may amount to what Harriet Hall, MD (a frequent contributor to the Science Based Medicine blog) has called “Tooth Fairy Science” [noted here]. This seeks research explanations for effects before establishing the prior probability for the existence of those effects; or, as she explains this…

“You could measure how much money the Tooth Fairy leaves under the pillow, whether she leaves more cash for the first or last tooth, whether the payoff is greater if you leave the tooth in a plastic baggie versus wrapped in Kleenex. You can get all kinds of good data that is reproducible and statistically significant. Yes, you have learned something. But you haven’t learned what you think you’ve learned, because you haven’t bothered to establish whether the Tooth Fairy really exists.”

This phenomenon may represent a larger domain that has been simply called “Fairy Tale Science,” which uses research (usually of questionable quality) to allege the validity of effects that may not truly exist; even though they may be believed to exist by millions of scientifically naïve minds. Fairy Tale scientists mistakenly think that if they can collect sufficient empirical research data, complete with statistically significant p-values, and the findings appear to be consistent with their hypothesis, then they have proven their hypothesis — such as of some treatment effect — to be true and valid. However, this Fairy Tale Science approach seeks verification of effects before establishing and taking into account whether or not those phenomena actually do exist in the first place; or, whether there are other, better explanations for the observed effects.

These are important concepts for evaluating all research in the pain field; but, especially when it comes to CAM modalities. Without belaboring the points further, and coming back to the question of foot reflexology, considering the lack of basic science — biological, physiological, or anatomical — to definitively explain how and why it might work, there would seem to be only minimal or no prior probability for its effects. Hence, it would require extensively robust and high-quality evidence from multiple trials demonstrating strongly favorable effects of reflexology for pain to overcome the lack of plausibility; and, judging from the relatively poor-quality and lackluster evidence to date, this has not been accomplished.
 

REFERENCES:
> Dechartres A, Trinquart L, Boutron I, Ravaud P. Influence of trial sample size on treatment effect estimates: meta-epidemiological study. BMJ. 2013;346:f2304 [
abstract here].
> Ernst E. Is reflexology an effective intervention? A systematic review of randomised controlled trials. Med J Aust. 2009 Sep 7;191(5):263-6.
> Ernst E, Posadzki P, Lee MS. Reflexology: an update of a systematic review of randomised clinical trials. Maturitas. 2011;68(2):116-120 [
abstract].
> Lee J, Han M, Chung Y, et al. Effects of foot reflexology on fatigue, sleep and pain: a systematic review and meta-analysis. J Korean Academy of Nursing. 2011;41(6):821-833 [
article here].
> Samuel CA, Ebenezer IS. Exploratory study on the efficacy of reflexology for pain threshold and tolerance using an ice-pain experiment and sham TENS control. Complementary Therapies in Clinical Practice. 2013(May);19(2):57–62 [
abstract].

eNotifications Don’t Miss Out. Stay Up-to-Date on Pain-Topics UPDATES!
Register [here] to receive a once-weekly e-Notification of new postings.

2 comments:

Liam Kitt said...

Dr Leavitt is incorrect in his analysis of the “tolerance data” in the paper by Samuels and Ebenezer. I have read the paper and they have analyzed the results correctly.

First they analysed the reliability of the pretreatment baseline data.

(Quoting from their paper) “The raw data for the pre-treatment baselines between sham TENS (control) and reflexology were 133.7 +/- 31.0s and 112.7 +/- 27.9s respectively. The inter-session reliability statistic for pre-treatment baselines between the two treatments was significantly correlated (Pearson product moment correlation coefficient: r = +0.70, df =15, p < 0.01) demonstrating good reliability for baseline data. Furthermore, the paired t-test confirmed that there were no significant differences between the baseline scores (p =3.55, n.s.).”

They then performed an ANOVA (Analysis of Variance) with repeated measures on treatment (as the participants were subject to both treatments in a crossover design) and time (as they measured effects on pain tolerance at 30 minute intervals for 120 min).

(Quoting from their paper) “Statistical analysis of the result showed significant main effect of treatment (F(1,14) = 5.1095, p < 0.05) and time (F(3,42) = 3.2505, p < 0.05), but no treatment x time interaction (F(3,42) = 1.6098, ns).”

The results of the ANOVA indicates:
(i)There was a significant main effect of treatment, which means that reflexology significantly increased pain threshold.
(ii) There were significant main effects of time, which means that the effects of treatment on this measure varied with time.
(iii) There was no treatment x time interactions, which means that the effects of reflexology on pain threshold did not diminish below that that of control measurements during the experimental period.

The ANOVA therefore shows that there is a clear-cut and significant difference of treatment between control treatment and reflexology which varies with time. This is clearly seen in the graphical representation of the data which shows that reflexology increases pain threshold compared with the Sham-TENS control data.

The authors then did, what is known as, post-hoc tests and found that reflexology significantly increased pain threshold at the 60 min, 90 min and 120 min intervals.

These authors have therefore used appropriate statistical tests and have shown indisputably that reflexology increases pain tolerance. I therefore find Dr Leavitt’s reasoning disingenuous as I am sure that he fully understands that his rudimentary analysis and subsequent rejection of the findings in this study are not appropriate.


SB. Leavitt, MA, PhD said...

Thank you, Laim Kitt, for your comments above. I always welcome constructive criticism of my analyses, since data in research reports can become confusing. I was not disputing the researchers’ ANOVA analyses and the significant differences between groups that were found, which I duly reported. However, the graphic depictions of their post hoc tests appear to tell a different sort of story, as I explained. If the error bars in those graphs had depicted Standard Deviations or Confidence Intervals, then it would have supported their contentions; however, since the researchers denoted those bars (in the graph caption) as Standard Errors (“se”) the alleged differences between groups are contradicted. Perhaps, they accidentally mislabeled the error bars, or there is something there that I’ve grossly misunderstood. Since they did not provide a complete display of the data behind those graphs in a table, it is impossible to know where the problem might lie.