Dose-Response Effects of Tai Chi and Physical Therapy Exercise Interventions in Symptomatic Knee Osteoarthritis

PM R. 2018 Jul;10(7):712-723. doi: 10.1016/j.pmrj.2018.01.003. Epub 2018 Jan 31.


Background: Therapeutic exercise is a currently recommended nonpharmacological treatment for knee osteoarthritis (KOA). The optimal treatment dose (frequency or duration) has not been determined.

Objective: To examine dose-response relationships, minimal effective dose, and baseline factors associated with the timing of response from 2 exercise interventions in KOA.

Design: Secondary analysis of a single-blind, randomized trial comparing 12-week Tai Chi and physical therapy exercise programs (Trial Registry #NCT01258985).

Setting: Urban tertiary care academic hospital PARTICIPANTS: A total of 182 participants with symptomatic KOA (mean age 61 years; BMI 32 kg/m2, 70% female; 55% white).

Methods: We defined dose as cumulative attendance-weeks of intervention, and treatment response as ≥20% and ≥50% improvement in pain and function. Using log-rank tests, we compared time-to-response between interventions, and used Cox regression to examine baseline factors associated with timing of response, including physical and psychosocial health, physical performance, outcome expectations, self-efficacy, and biomechanical factors.

Main outcome measures: Weekly Western Ontario and McMasters Osteoarthritis Index (WOMAC) pain (0-500) and function (0-1700) scores.

Results: Both interventions had an approximately linear dose-response effect resulting in a 9- to 11-point reduction in WOMAC pain and a 32- to 41-point improvement in function per attendance-week. There was no significant difference in overall time-to-response for pain and function between treatment groups. Median time-to-response for ≥20% improvement in pain and function was 2 attendance-weeks and for ≥50% improvement was 4-5 attendance-weeks. On multivariable models, outcome expectations were independently associated with incident function response (hazard ratio = 1.47, 95% confidence interval 1.004-2.14).

Conclusions: Both interventions have approximately linear dose-dependent effects on pain and function; their minimum effective doses range from 2-5 weeks; and patient perceived benefits of exercise influence the timing of response in KOA. These results may help clinicians to optimize patient-centered exercise treatments and better manage patient expectations.

Level of evidence: II.

Publication types

  • Randomized Controlled Trial
  • Research Support, N.I.H., Extramural

MeSH terms

  • Adult
  • Exercise Therapy / methods*
  • Exercise Tolerance / physiology*
  • Female
  • Follow-Up Studies
  • Humans
  • Knee Joint / physiopathology
  • Male
  • Middle Aged
  • Osteoarthritis, Knee / physiopathology
  • Osteoarthritis, Knee / rehabilitation*
  • Quality of Life
  • Range of Motion, Articular / physiology*
  • Self Efficacy*
  • Self Report
  • Single-Blind Method
  • Tai Ji / methods*

Associated data