Exercise and knee osteoarthritis: benefit or hazard?

Quality of evidence

MEDLINE, EMBASE, DARE, ACP Journal Club, and Cochrane databases were searched from registry inception to January 2009 using MeSH headings or text words, including osteoarthritis, arthritis and knee and exercise, physical training, and run. Reference lists from retrieved articles, citation listings when available, and related articles suggested in PubMed were also evaluated. As more vigorous forms of exercise interventions are employed in patients without joint symptoms, there was no expectation that there could be adequate randomization or blinding; therefore, level II evidence was deemed acceptable for studies of patients without OA. For studies of patients with established arthritis, only level I evidence that included systematic reviews was selected, as there was outstanding high-level evidence for this cohort. In all cases, the best evidence is presented in this review.

Selection criteria differed for studies of patients with and without established knee OA. Because studies of patients without pre-existing disease were of poorer quality, accepted criteria were much more liberal for this group (Table 1). Conclusions drawn for patients without established OA are therefore less reliable than those for patients with existing disease.

Definition and diagnosis of knee OA

Although the onset of knee OA probably involves the entire joint, loss of hyaline cartilage seems to be the signature event. The presence of inflammation is inconsistent, and is not clearly causative. Diagnosis is both clinical and radiological (Box 1); however, clinical features form the basis of diagnosis, with x-ray investigations being helpful to confirm diagnosis and exclude other possible conditions.

There is considerable discordance between joint symptoms and x-ray findings in knee OA.1,6–8 Patients with radiographic evidence of OA were found to have knee pain between 15% and 81% of the time in a recent systematic review.8 Imaging techniques are also important. Patellofemoral views can improve the likelihood of predicting the cause of knee pain from 10% to 50%.9 Conventional weight-bearing anterior-posterior views done for patients who have acute pain often overestimate joint space narrowing, as pain obliges the patient to maintain the joint in slight flexion.10 On the other hand, early painful OA might not necessarily be accompanied by radiographic changes. This can happen for several reasons:

• Pain fibres are not present in cartilage, so pain might not be perceived until the periosteum, joint capsule, peripheral meniscus, or synovium become stimulated.

• Comorbidities, such as surrounding muscle weakness and loss of proprioception, can both be the cause and the result of OA. Subsequent disturbed joint function results in pain.1

Clearly there is substantial evidence that, in the absence of a clinical correlation, x-ray findings have little relevance in understanding the cause of knee pain.

Etiology

Various proposed causes of and risk factors for knee OA are summarized in Table 2.11–25 For the purposes of this review, factors related to the risks and benefits of exercise are considered. The first consideration with respect to exercise is whether “wear and tear” resulting from repetitive use of articular cartilage is the primary driver for degenerative change. If this were the case, exercise would lead to progressive OA. According to the evidence presented here, this does not seem to be the case. Alternatively, Schrier, in a systematic review,24 has proposed that muscle dysfunction and weakness resulting from injury, inadequate rehabilitation, overuse, and inactivity is the primary driver for OA. In normal knees, the muscles absorb most of the forces presented to the joint. Articular cartilage does not absorb the remaining force, but redistributes it to bone. Abnormal force distribution on cartilage caused by injury, malalignment, meniscectomy, or muscle dysfunction eventually leads to cartilage damage; loss of this redistribution leads to protective bony sclerosis. The hardened bone, being less resilient, leads to further cartilage damage, facilitating osteoarthritic change.

It seems possible that a normal joint, experiencing normal forces, is well-protected when subjected to even vigorous exercise, particularly if there is opportunity for adaptation and regeneration through incremental muscle training and injury avoidance. Trained marathon runners have been found to have normal post-race magnetic resonance imaging findings, while beginner runners show abnormalities after the same distance.26 In contrast, abnormal forces on the joint resulting from trauma,27 poor proprioception,20 joint misalignment,22 or muscle dysfunction and weakness24 might provoke osteoarthritic changes after exercise.

What is the role of exercise in causing knee OA?

A 60-year-old man has been running 30 km/wk for 35 years. He has no history of noteworthy illness and has never had a sports- or running-related injury. He has been told that he is wearing out his knees and that he will get arthritis if he continues to run, but he wishes to remain active as he gets older. He asks your advice.

Evidence for the influence of exercise on knee OA onset is all level II, owing to problems with compliance and blinding in imposing this type of intervention. Studies are observational and are done on self-selected populations. There is no externally imposed intervention. These studies do, however, provide an opportunity to evaluate quite vigorous levels of exercise, and many of them are of prospective cohorts followed for very long periods of time.

While presence of knee OA is often the end point in these studies, there is generally a poor correlation between x-ray findings and symptoms. A diagnosis of knee OA is really only a surrogate for lower extremity pain and disability, which are major components of reduced quality of life in seniors. The process advances with age equally in those who exercise and those who do not.28 The best prospective cohort studies12,29–35 show no increase in rate of progression of knee OA in individuals who exercise compared with those who do not. Many of these populations are followed for long periods of time (Table 312,13,17,21,22, 27–44). Only one cohort study17 shows increased OA in individuals who exercise, and only in those with a history of “heavy physical activity.” More important, there is convincing evidence—as well as lack of dissenting opinion—that increased levels of exercise lower the incidence of musculoskeletal disability, reduce pain, and increase functional capacity.28,31,37 In fact, Stanford University’s runners’ study28 shows that after 21 years all-cause morbidity is further delayed toward the end of life and that, so far, mortality is also being delayed in runners compared with controls.

Studies of lesser quality, usually case-control studies or cross-sectional surveys, have been conducted, which show similar results with high levels of exercise.36,40,42 Some of these suggest that athletes competing at high levels of activity when they are young are increasingly susceptible to OA later in life.13,44 Elite athletes have shown increased knee OA in some studies.27,39,41,43 Some authors cite trauma as a risk factor.21,22,27 Certain sports, such as soccer and wrestling, increase the risk.35 Genu varum was associated with knee OA in one paper.22 Occupational stresses, such as repetitive kneeling and stair climbing, can increase risk.19 Surprisingly, marathon running does not seem to induce changes in joints or increase the risk of OA in most studies.40,45,46 A large case-control study comparing swimmers with runners was unable to show a difference in lower-extremity pain or arthritis surgery associated with either accumulated mileage or number of years spent running.36

To summarize this literature:

• The best evidence suggests that exercise, at least at moderate levels, does not accelerate development of knee OA. Running seems to be particularly safe.

• There might be increased risk of OA with competitive sports participation, particularly early in life, and with competition at an elite level; however, the presence of OA does not lead to increased disability.

• Risk of OA might be increased in the presence of obesity, trauma, occupational stress, and alignment problems of the lower extremities.

• There is evidence for reduction in lower-extremity disability and all-cause disability in self-selected runners compared with controls.

• There is some evidence for prolongation of lifespan in self-selected runners.

Case resolution

This man can be reassured that, as a self-selected runner, he will not develop accelerated knee OA as a result of his activity, and that his risk of disability might even be minimized as he ages because of his level of exercise. Although there is no evidence that a physician-imposed recommendation for the same level of activity in a sedentary person would be risk free, it is reassuring to note that activity at a moderate level with avoidance of extreme sports and trauma does not seem to increase the risk of knee OA. Knee x-ray scans would not be helpful for this man, as changes do not predict future appearance of clinical symptoms.

In the presence of knee OA, what is the effect of exercise on physical function, pain, and disability?

A 55-year-old woman has been experiencing increasing knee pain with physical activity for the past several years. She has been told by her physician that she presents early stages of knee arthritis on x-ray scans, and she occasionally uses over-the-counter anti-inflammatory medications for relief. She is slightly overweight and finds that she cannot lose weight by dieting. She is sedentary and is afraid to start exercising as an aid to weight loss because she is afraid that this will make her knee arthritis worse as she becomes older. She asks your advice.

Review of the literature addressing the effects of exercise on established symptomatic knee OA yields numerous good-quality level I studies in the form of Cochrane reviews and systematic reviews (Table 447–55). There is, in fact, a systematic review evaluating available systematic reviews.50 Blinding is still a problem in some of the more vigorous interventions. There are also small problems with randomization, intention-to-treat analysis, compliance, and heterogeneity in the evaluation of these studies; however, all included studies are randomized controlled trials.

Interventions examined include progressive resistance training,49 quadriceps strengthening,52 aquatic exercise,47 land-based exercise,48 walking,53 intensity of life activities,51 aerobic training,52 and all physical modalities.50 Long-term effects of exercise interventions have been evaluated.54 Factors influencing prognosis for OA progression have also been identified.55 The following conclusions can be drawn from this literature:

• These studies pertain only to the effects of interventions on symptoms and degree of disability in patients previously diagnosed with knee OA. No clinical or x-ray evaluations were done.

• The levels of exercise intervention were generally low or moderate.

• There is demonstrated benefit for sedentary people, provided progressive structured activity is provided.51

• There is little correlation between severity of OA symptoms at onset of activity and degree of benefit.55 here is probably an advantage of dynamic over static activity.51

• Obesity is a prevalent comorbidity. Studies varied as to whether obese subjects benefited from exercise interventions,55 but the most inclusive study suggested benefit.50 Body fat reduction by exercise and diet, rather than weight reduction alone, was effective in reducing OA symptoms in a small controlled study.56 Therefore, it is suggested that weight loss need not be a prerequisite for OA improvement in overweight individuals. None of the exercise modalities demonstrated any harm in obese subjects.

• Most of the benefits of exercise were measured for a short term.47,48,53 Long-term benefit was not demonstrated unless repeated “booster sessions” were provided.54

• Exercise interventions all had small to moderate beneficial effects on pain. There were no dissenting studies.

• All exercise interventions, with the exception of progressive resistance training, had a small to moderate beneficial effect on disability.

• The benefits of land-based exercises can be similar to the relief obtained from nonsteroidal anti-inflammatory drugs.48 Generally, nonpharmacologic approaches to OA treatment are as effective as pharmacologic approaches.57

• There is no indication that low to moderate exercise intervention causes increased pain or disability in knee OA.

Guidelines from the American College of Rheumatology58 and the European League Against Rheumatism,59 although generated in 2000, are consistent in recommending aerobic exercise, muscle strengthening, and preservation of joint mobility as central components of the nonpharmacologic approach to the treatment of knee OA. The 2007 recommendations of the Osteoarthritis Research Society International57 are more inclusive of the current literature and again stress the value of exercise, whether water-based, aerobic, muscle strengthening, or physiotherapy-based. Further, these guidelines stress that there are no statistical differences in the benefits derived from pharmacologic and nonpharmacologic modalities in the treatment of knee OA.57 Despite these findings, DeHaan et al,60 in a Canadian study at a teaching clinic, were rarely able to find documentation of the use of nonpharmacologic therapy. However, based on the evidence, judicious use of exercise is a compelling modality in the treatment of knee OA, and this intervention incurs little risk.

Case resolution

This woman can be reassured that gradually increasing exercise to a moderate level as a means of losing weight is unlikely to make her knee symptoms worse. She can also be told that there is a possibility that exercise can improve her knee symptoms as well. A plan for continuing long-term exercise will be important to both facilitate and maintain her weight loss and possibly reduce her knee symptoms. It is reasonable to suggest that, even if she achieves minimal weight loss, continued exercise will minimize both pain and disability. If she chooses land-based exercise, she might expect improvement in symptoms and disability similar to benefits obtained from nonsteroidal anti-inflammatory drugs, without the side effects. She might consider stopping the medication or switching to acetaminophen.

Conclusion

Individuals without knee OA who opt to exercise will not have increased progression of joint degeneration as a result of their increased physical activity; indeed, they can expect reductions in knee pain and all-cause disability as the years progress. Vigorous sports and activity leading to trauma should be avoided. Externally imposed recommendations for exercise will not necessarily produce the same results, as these studies have not yet been done.

Persons with knee OA who exercise to a moderate level can expect reduction in both knee pain and disability for the duration of their intervention. A long-term exercise program is unlikely to be harmful, but interventions of a longer duration have not been adequately studied. There is outstanding evidence for the benefit of exercise therapy in knee OA and some indication that it is underused as a treatment modality.

Levels of evidence

Level I: At least one properly conducted randomized controlled trial, systematic review, or meta-analysis

Level II: Other comparison trials, non-randomized, cohort, case-control, or epidemiologic studies, and preferably more than one study

Level III: Expert opinion or consensus statements