Blinded studies show far lower efficacy than unblinded studies


Published 1/1/2016
Terry Stanton

Taking a Second Look at Effectiveness of Viscosupplementation

When the AAOS first issued a clinical practice guideline (CPG) on nonarthroplasty treatment of knee osteoarthritis (OA) in 2009, the evidence on the effectiveness of viscosupplementation (hyaluronic acid [HA] injections) as a treatment option was inconclusive. When the CPG was reviewed and reissued in 2013, the evidence—according to the work group—was undeniable: "We cannot recommend using hyaluronic acid for patients with symptomatic osteoarthritis of the knee," they wrote.

Since then, a spate of new studies have tried to call that conclusion into question. However, a systematic review of studies on the impact of HA injections for knee OA, published recently in The Journal of Bone & Joint Surgery, found that when only the best-quality evidence is considered, such viscosupplementation did not demonstrate a clinically important benefit over placebo.

The study, explained lead author David Jevsevar, MD, MBA, is the first to report results on HA using best-evidence techniques. This approach follows principles developed by education researcher Robert E. Slavin, PhD, who expressed concern about the use of all data and trials in performing a meta-analysis.

"When all data are used," Dr. Jevsevar and fellow authors noted in their article, "the data parameters from poorly performed, and on occasion non–peer-reviewed, studies attain equivalency to those from peer-reviewed studies performed with greater internal validity and patient validity."

A meta-analysis, explained Dr. Jevsevar, typically looks at all published data, and sometimes unpublished data as well. "This generates bigger numbers, and the thought was that bigger numbers could lead to a better conclusion. Dr. Slavin suggested not doing the analysis of all data, but just the best literature. That enables researchers to compare apples with apples. If both good studies and poor studies are included, it's like comparing apples with oranges."

In their meta-analysis, said Dr. Jevsevar, "We took this Slavin best-evidence approach. We included the concept of blinding, with the physician and the patient unaware of the treatment the patient was getting."

The topic of viscosupplementation was selected, according the Dr. Jevsevar, primarily because it has a "huge" literature base. "Some people will look at this and say that we wrote it to say that HA doesn't work. The conclusion of the paper is that, based on an analysis of blinded and appropriate studies, HA is not effective. The emphasis is on performing an analysis using the best evidence and showing the difference that blinding makes."

Clinical significance: The key criterion
In selecting the papers that would be included in the analysis, the authors scanned databases for randomized, controlled trials that compared HA with a control treatment and had a minimum of 30 patients per subgroup. For consideration, each article had to include VAS or WOMAC pain, WOMAC function, and/or WOMAC stiffness as outcomes, because the minimal clinical important difference (MID) has been established for these instruments.

MID, Dr. Jevsevar explained, is a concept the Academy applies in developing CPGs. (See "The Importance of Clinical Significance in AAOS Clinical Practice Guidelines," AAOS Now, May 2013.) "MID states that things can be statistically different, but the difference may not be clinically relevant," he said. "For example, a study examining two different treatments for tibia fractures may find that healing with one treatment one was statistically faster than the healing with the other, but the time difference was less than a day. Is that clinically important? The answer would be no. MID looks at whether the change that was measured is clinically relevant. It assumes that a statistically significant answer is also clinically relevant."

Most previous systematic reviews evaluating the efficacy of HA have relied strictly on statistical significance as the determinant, without considering the clinical relevance of the difference.

In this study, of 628 abstracts originally identified by a literature search, 545 did not meet the inclusion criteria. After the remaining 83 were given a full text review, 19 articles, representing 4,485 patients, were selected. Fourteen of the trials compared HA with placebo (sham treatment), two compared HA with conventional (usual) care, and three paired HA with an additional active treatment and compared the results with those in a control group that received active treatment alone.

In the sham-controlled trials, the authors reported that the HA group had significantly better pain scores. "However," they wrote, "the average treatment effect was only 29 percent of the MID. It is unlikely that an appreciable number of patients received a clinically important benefit, as the average treatment effect was less than 50 percent of the MID.

"In the trials that compared HA plus an additional active treatment with that active treatment alone, the effect size was still only about one-half (51 percent) of the MID. The treatment assignment in these trials was not blinded, as the treatment group received two interventions (HA plus additional treatment) whereas the control group received only one intervention (the same additional treatment as in the treatment group, with no HA). However, the fact that the control group received an active treatment instead of a placebo may have tempered some of the placebo effect.

"The final group of trials that compared HA with usual (appropriate) care provided no control for the placebo effect. The treatment effect was far greater in these trials than in the previous two groups of trials, with the overall effect size of 1.52 MID units being 5.2 times greater than the effect size of 0.29 MID units in the double-blinded trials."

Other groups that have published in this area have said that a 20 percent improvement is sufficient, according to Dr. Jevsevar. His group specifically used two studies that looked at clinical significance in patients with knee OA. These studies estimated treatment effects of 0.5 to 1.0 MID units may be beneficial to an appreciable number of patients. They concluded that, as the treatment effect drops below 0.5 units, it would be unlikely that an appreciable number of patients will show a clinically important benefit.

Another consideration was the length of follow-up. Many studies report only on early follow-up. "We know that almost any intra-articular injection may give people relief for a short period—probably 2 to 4 weeks," noted Dr. Jevsevar, "although the literature doesn't unequivocally show that.

"With this best evidence analysis and appropriate blinding applied to studies, we found that intraarticular HA does not have a clinically significant effect," said Dr. Jevsevar of the overall results. "When appropriate blinding and appropriate best evidence were used, the effect size was not large enough to show a benefit."

Debate continues
HA remains in widespread use. A recent statement from the American Medical Society for Sports Medicine (AMSSM), which represents nonsurgical sports medicine physicians "recommends the use of HA for appropriate patients with knee osteoarthritis." That conclusion was based on a literature review that used the Outcome Measures in Rheumatoid Arthritis Clinical Trials–Osteoarthritis Research Society International (OMERACT-OARSI) and several studies excluded by Dr. Jevsevar's group.

For example, the AMSSM review included studies with fewer than 30 patients per subgroup as well as those with "no comparison of interest"—meaning the comparisons performed were between varieties of HA, and not HA versus placebo or other treatment.

"If you look at just statistical significance, then HA appears to work, although its effect size is pretty small," Dr. Jevsevar said. "As soon as you apply any reasonable bar of clinical significance, it doesn't work. Many clinicians will argue that they have used HA for several years, and 'it works in our patients.' But just because we perform an injection and the patient doesn't come back for 6 months or a year, that doesn't mean the treatment worked.

"First of all, knee arthritis doesn't hurt the same every day, so the patient may have entered a phase where it wasn't as bothersome," continued Dr. Jevsevar. Or, the patient may not want other treatments. The patient may not want a knee replacement or hard discussions and may not return. But unless the study follows patients through various time intervals, it won't show how well HA injections really work for them."

The co-authors of the article are Gregory A. Brown, MD, PhD, and two members of the Academy's Research and Scientific Affairs Department—Patrick Donnelly, MA, research analyst; and Deborah S. Cummins, PhD, department director. The authors' disclosure information can be accessed at

Terry Stanton is a senior science writer for AAOS Now. He can be reached at

Bottom Line

  • A meta-analysis of "best-evidence" studies found that hyaluronic acid injections do not demonstrate a clinical benefit over placebo in the treatment of knee osteoarthritis.
  • A statistically significant difference between treatments may not be clinically relevant, because the difference does not meet the minimal importance difference (MID) for clinical relevance; a statistically significant difference in length of stay, for example, may be a few hours, but that would not be clinically relevant to the patient.
  • Despite ongoing use of HA in the treatment of knee OA, the effect size of the injections is not large enough to show a benefit when appropriate blinding and appropriate best evidence are applied to trials involving HA injections.


  1. Jevsevar DS, Brown GA, Cummins DS, Donnelly P. Viscosupplementation for Osteoarthritis of the Knee: A Systematic Review of the Evidence. J Bone Joint Surg Am, 2015 Dec 16; 97 (24): 2047 -2060.
  2. Slavin RE. Best-evidence synthesis: an alternative to meta-analytic and traditional reviews. Educ Res. 1986;15(9):5–11.
  3. Rutjes AW, Jüni P, da Costa BR, Trelle S, Nüesch E, Reichenbach S. Viscosupplementation for osteoarthritis of the knee: a systematic review and meta-analysis. Ann Intern Med. 2012 Aug 7;157(3):180–191.
  4. Colen S, van den Bekerom MP, Mulier M, Haverkamp D. Hyaluronic acid in the treatment of knee osteoarthritis: a systematic review and meta-analysis with emphasis on the efficacy of different products. BioDrugs. 2012 Aug 1;26(4):257–268
  5. Jevsevar D. The Importance of Clinical Significance in AAOS Clinical Practice Guidelines, AAOS Now, May 2013.