Reading a Massage Study: Outcomes, Effect Sizes and Short-Term vs Lasting Benefit

Reports on massage often say "significant improvement" or give a number like "SMD -0.75". This page explains those terms using real examples from the reviews we read. It is a reading guide, not medical advice; for what applies to you, ask your own clinician, and do not use massage to postpone seeing one [1].

Step 1: What was measured?

Most massage trials measure pain, function, anxiety or sleep by asking people to rate themselves. Those are patient-reported outcomes. They matter, because pain and comfort are what people care about, but they are open to expectation effects when nobody is blinded; see How Massage Is Studied. Ask: did the trial measure what I care about, and is it a self-rating or a measurement?

Step 2: Compared with what?

A result "better than nothing" is weaker than "better than a sham" or "better than exercise". The low-back Cochrane review separated inactive controls (sham therapy, waiting list, or no treatment) from active ones [2]. In it, massage did better than inactive controls for pain in sub-acute and chronic back pain (SMD -0.75, 95% CI -0.90 to -0.60; 761 participants; 7 studies), but the effect was short-term [2].

Step 3: How big is the effect?

Many reviews use a standardized mean difference (SMD) so different pain scales can be combined. A common rule of thumb from the Cochrane Handbook is that 0.2 represents a small effect, 0.5 a moderate effect and 0.8 a large effect, though it notes variations exist and these rules came mainly from the social sciences [3]. By that rule, an SMD of -0.75 is between moderate and large, but the review still had very little confidence in the evidence because of its quality [2].

Look at the confidence interval (CI): the range in which the true effect probably lies. A narrow range around a large number is reassuring. A wide range, or one that crosses zero, is not. Also look at how many people the number rests on. The same review reported a very large effect for acute back pain (SMD -1.24, 95% CI -1.85 to -0.64), but that came from one study of 51 participants [2].

Step 4: Is the change big enough to matter?

A "statistically significant" result can be too small to notice. For low back pain, an international panel proposed that a change of 15 points on a 0 to 100 pain scale, or 2 on a 0 to 10 scale, is a meaningful improvement for an individual, and that a 30% change from baseline is a useful threshold [4]. They called these values a starting point, not the final answer [4].

Be careful when comparing a trial's average difference with those numbers. A 2026 paper argued that thresholds based on within-person change are often inappropriately applied to between-group effects in trials and meta-analyses [5]. An example: in the Cochrane neck-pain review, pain was a mean of 20.55 points with placebo and improved by 3.43 points with massage on a 0 to 100 scale, and the range of uncertainty ran from 8.16 points better to 1.29 worse [6]. That small average and wide range is why the review described little to no difference against placebo [6].

Step 5: Short-term or lasting?

Many positive findings are short-term. The low-back review found pain improvements with massage only in the short-term follow-up [2]. A large review of non-drug treatments for chronic pain noted that many trials were small and that evidence beyond 12 months after treatment was sparse [7]. A trial that measures pain right after a session says little about next month. For neck and shoulder pain, NCCIH says the benefits may only last for a short time [1].

Step 6: How certain is the evidence?

Reviews often grade certainty (high, moderate, low, very low). The low-back review judged the evidence to be "low" to "very low" [2]; the American College of Physicians evidence review rated massage for chronic low back pain as low to moderate, with small to moderate and generally short-term pain benefits [8]. Also check dose: the neck-pain review found a clinically important difference in a high-dose subgroup (eight or more sessions over four weeks, 30 minutes or longer) [6]. That is a finding about what trials gave, not a recommendation of how much to have.

A quick checklist

• Who was studied (age, condition, how long they had it)?
• What exactly was done (hands, device, style, minutes)?
• What was the comparison?
• Which outcome was main, and was it reported honestly?
• How many people, and how many dropped out?
• Is the effect large enough to matter, and does it last?
• Who funded it? In the low-back review the majority of trials had not-for-profit funding [2].

Condition-specific findings are on Massage Evidence by Goal. For a broader primer see Understanding Evidence in Complementary Health.