Physical therapy better than steroid injections for knee osteoarthritis after 1 year

Référence

Deyle GD, Allen CS, Allison SC, et al. Physical therapy versus glucocorticoid injection for osteoarthritis of the knee. N Engl J Med 2020;382(15):1420-1429.

Sommaire

These researchers identified patients, 38 years and older, in the Military Health System, including active duty military and retirees, who met ACR criteria for osteoarthritis of the knee (including radiographic evidence). Patients who had received either a glucocorticoid injection in the knee or PT in the previous year were excluded. The authors randomized 156 participants to receive either up to 3 injections of 40 mg triamcinolone plus 7 mL of lidocaine 1% over the course of a year or up to 8 PT sessions in the first 4 to 6 weeks with additional sessions as needed (ultimately, a mean of 11.8 PT visits). Outcomes included a series of standard symptom scales, such as the Western Ontario and McMaster Universities Arthritis Index (WOMAC), assessed in the knee with worse symptoms. The trial was powered to detect a 12% difference in WOMAC scores, which is thought to be the minimal clinically important difference. Analysis was by the intention-to-treat principle, and groups were balanced at the beginning of the study. This was an open-label trial, but the outcomes were assessed by investigators who were masked to treatment assignment. The mean age of patients was 56 years, 48% were women, and the average baseline WOMAC score was similar between groups. During the course of the year, both groups saw a rapid and similar decline in WOMAC scores during the first month. The WOMAC scores continued to decline for the PT group, but plateaued for the injection group. The primary outcome was change in the WOMAC score at 1 year, which decreased more in the PT group than in the injection group (mean scores: 37 points vs 55.8 points; P < .05). Patients in the injection group received a mean of 2.6 injections. There were some crossovers: 9% of patients in the PT group received a glucocorticoid injection and 18% in the injection group received some PT. A secondary measure, the Global Rating of Change score, also improved slightly more in the PT group, and costs for knee care were similar between groups. The study is limited by the open-label design, which could lead to a Hawthorne effect for those in the PT group and a placebo effect for those in the injection group. The rapid and large improvement in the first month for both groups is somewhat surprising, especially for the PT group, with relatively little further improvement seen in the rest of the year, which suggests that regression to the mean may also have contributed to the observed improvement.