Deyle et al (2020)

Steroid Injections vs Physical Therapy for Knee OA

Physical therapy reduced pain and increased function more than a steroid injection

Different modalities like manual therapy, exercises, and education, were included


Knee osteoarthritis (OA) is a leading cause of disability. A glucocorticoid injection (GCI) is commonly used as a first-line treatment, yet conflicting reports exist. Clinical practice guidelines often differ in their recommendations for GCI’s.

Physical therapy treatments might confer short-term and long-term relief of common symptoms and functional disabilities. Adding to this, a decrease in pain medications would be an interesting byproduct. Interestingly, a large database analysis revealed GCI’s are four times more likely than receiving physical therapy treatment before total knee replacement.

The current study compared the effectiveness of GCI’s to physical therapy.



Patients included were beneficiaries of the Military Health System and were active-duty or retired, or family members. Criteria to be met were those of the American College of Rheumatology, together with radiographic evidence of OA assessed as Kellgren-Lawrence grade 1-4. Patients that received either GCI or physical therapy treatment for their knee in the previous twelve months were excluded. Patients were assigned in a 1:1 ratio to either physical therapy or GCI. Before randomization, patients were provided with education, based on current guidelines, addressing the relationship between OA, physical activity, nutrition, and obesity.

GCIs were administered by skilled physicians.

Patients were reassessed after four and nine months, including the appropriateness of additional injections (no more than three). The physical therapy treatment included exercises, joint mobilizations, clinical reasoning regarding priorities, dosing, and progression. A typical session included manual techniques preceding exercises. Subjects underwent up to eight treatment sessions over the initial four to six weeks and could opt for an additional one to three sessions at four and nine months if agreed upon with the physical therapist.

The primary outcome measure was the Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) at one year. Secondary outcome measures were Global Rating of Change scale (GRC), Timed Up and Go (TUG), and the Alternate Step Test (AST). The WOMAC MCID was a 12-16% improvement from baseline.

A priori power analysis revealed 138 subjects were needed to detect interaction effects for time and group.



The study enrolled 156 patients; male to female ratio was practically equal and the average BMI was 31.5. Patients in the GCI group received a mean of 2.6 injections. Patients in the physical therapy group received a mean of 11.8 treatment sessions. Nine percent of the physical therapy group also received an injection. Conversely, 18% of the GCI group also received physical therapy treatment.

The mean WOMAC score at baseline was around 93/240 for both groups. After one year, the GCI group dropped to 55.8/240 and the physical therapy group to 37.0 — the lower the better.

Table 2 deyle et al knee oa
From: NEJM, Deyle 2020

The GRC scale noted “quite a bit better” (+5) in the physical therapy group and “moderately better” (+4) in the GCI group. Additionally, patients receiving physical therapy performed better on the TUG and AST.


Questions and Thoughts

A huge ‘win’ for physical therapy it seems. Let’s get into the details. First things first, not every patient experienced dramatic improvement. About 10% did not improve clinically meaningful in the physical therapy group, compared to 25% in the injection group.

The ones that did improve, improved quite drastically in the first four weeks, compared to follow-up. This could be explained by provider contact for their complaint, the educational sessions, regression to the mean, or a combination. Contact time is one of the first limitations of the study. Patients in the physical therapy group had notably more appointments, which could alter results.

Also, is this a fair comparison regarding timeframes? Without completely bashing glucocorticoid injections, if they’re used we know they won’t last for more than a few weeks. Let alone months (ref.).


Talk nerdy to me

A thing worth noting is the large confidence interval (CI), suggesting uncertainty, around the between-group difference (18.8, 95% CI 5.0 – 32.6).

The physical therapy group received a number of different modalities including manual therapy, exercises, and education. In contrast, the American College of Rheumatology/Arthritis Foundation recommends against manual therapy. That aside, multiple interventions make it hard to assess the ‘working ingredient’.
Research shows that injection benefits are short-lived. One might say a twelve-month follow-up is unreasonable; however, a second and third injection was an option.

A final limitation of this study is that it was set up to be a multicenter trial. Not everything went according to plan since one of the two centers provided only four eligible patients.

Secondary analysis showed superior effects for physical therapy. Yet, no definitive statements can be made since this was not the goal of the study, as the authors correctly noted. Secondary outcomes should be taken with a few grains of salt since studies are not powered for this.

This is a high-quality and much-needed study. We should see more of these getting published, although replicating is hard since ‘guideline-based multimodal physical therapy’ is up for interpretation by the therapist.


Take home messages

  • Physical therapy treatment could be superior to glucocorticoid injection therapy for the short, middle, and long-term
  • Future studies should warrant more certainty regarding the specific-, financial-, and preventative effects of physical therapy
  • Physical therapy is a worthy first-line intervention for knee OA



Deyle, G. D., Allen, C. S., Allison, S. C., Gill, N. W., Hando, B. R., Petersen, E. J., … & Rhon, D. I. (2020). Physical therapy versus glucocorticoid injection for osteoarthritis of the knee. New England Journal of Medicine382(15), 1420-1429.



Watch this FREE video lecture on Nutrition & Central Sensitisation by Europe’s #1 chronic pain researcher Jo Nijs. Which food patients should avoid will probably surprise you!

CS Diet
Download our FREE app