There is inadequate evidence to determine the effectiveness of nonpharmacological and nonsurgical interventions for hand osteoarthritis: an overview of high-quality systematic reviews.

Authors: Moe RH (1) , Kjeken I (1) , Uhlig T (1) , Hagen KB (2)
(1) Diakonhjemmet Hospital, National Resource Centre for Rehabilitation in Rheumatology (2) Institute of Nursing and Health Sciences, Faculty of Medicine, University of Oslo
Source: Phys Ther. 2009 Dec;89(12):1363-70
DOI: 10.2522/ptj.20080398 Publication date: 2009 Dec E-Publication date: Oct. 22, 2009 Availability: full text Copyright: © 2009 American Physical Therapy Association
Language: English Countries: Not specified Location: Not specified Correspondence address: Moe RH :


Article abstract


Patients with hand osteoarthritis are commonly treated by health care professionals (allied to medicine). Practice should be informed by updated evidence from systematic reviews of randomized controlled trials.


The purpose of this overview is to summarize the evidence from systematic reviews of the effectiveness of nonpharmacological and nonsurgical interventions for patients with hand osteoarthritis.


Systematic reviews published between January 2000 and October 2008 were identified by a comprehensive literature search.


Two reviewers independently selected reviews for inclusion, assessed their methodological quality, and extracted and synthesized data according to predefined criteria. Four systematic reviews finally were included. Based on single randomized controlled trials, there is some evidence of the effect of pain relief from topical capsaicin compared with placebo and for favorable functional outcomes for exercise and education compared with osteoarthritis information alone.


In overviews, results are dependent on available systematic reviews. They are important tools to guide choice of interventions and locate areas where more research is needed, but they might not be useful for deciding specifically how interventions should be carried out.


There currently is insufficient high-quality evidence regarding nonpharmacological and nonsurgical interventions for hand osteoarthritis. Considering the limited research evidence and the prevalence and impact of the disease, there is an urgent need for more trials of nonpharmacological and nonsurgical interventions for hand osteoarthritis.

Article content

Osteoarthritis (OA) is a chronic joint condition characterized by loss of articular cartilage and new bone formation and is associated with pain, functional disability, and impaired quality of life.1 Most people with OA are female, and the condition increases in prevalence with age. Symptomatic OA in at least one hand joint occurs in about 20% of the population, and the majority of people over 55 years of age show radiographic hand OA.2,3 The occurrence of OA may vary depending upon the population studied and the diagnostic methods used.4 In a recent population-based study in Norway, the prevalence of self-reported hand OA in women between 24 and 76 years of age was 5.8%, whereas the corresponding estimate for men was 2.5%.5

Despite its growing burden on society, OA remains a poorly understood disease. At present, no disease-modifying interventions are available,6 and current treatment of OA, therefore, is aimed mainly at alleviating symptoms and includes pharmacological approaches, physical therapy, exercises, braces and orthoses, weight reduction, and surgery.4 The European League Against Rheumatism (EULAR) endorses recommendations of a combination of pharmacological and nonpharmacological care in treating people with hand OA.7 However, although evidence of the effectiveness of education and exercise for reducing pain and improving physical functioning is convincing regarding knee OA,1 such evidence still is very uncertain for hand OA7 and hip OA.8

Decisions regarding the provision of health care increasingly are based on the available evidence from high-quality clinical research. Patients, health care professionals, and researchers need information about the effectiveness of interventions so that they can improve self-management strategies and clinical practice and set research priorities. Decisions regarding health care reimbursement also are increasingly evidence based. Thus, purchasing organizations and policy makers in health care will demand reliable information on the effectiveness of interventions.

Conclusions based on a systematic review of randomized controlled trials (RCTs) are considered to provide the highest level of evidence about the effectiveness of an intervention. Based on a review of literature up to 2001, Chard and Dieppe concluded that nonpharmaceutical therapies for OA have not been researched enough for us to understand their potential benefit.9 The aim of this overview is to summarize currently available evidence from systematic reviews on the effectiveness of nonpharmacological and nonsurgical interventions for patients with hand OA.


Criteria for Including Reviews

We included systematic reviews with the primary aim of investigating the effects of nonpharmacological and nonsurgical interventions for hand OA published in the English, Dutch, or Scandinavian language. More specifically, the following inclusion criteria were used:

  • People with hand OA: Diagnosis according to the American College of Rheumatology criteria10 or other acceptable criteria. Reviews including people with OA in other joints or various rheumatic diagnoses were accepted only if results for hand OA could be extracted separately.

  • Interventions: All types of nonpharmacological and nonsurgical interventions. Excluded were interventions such as gene therapy, all types of invasive interventions (eg, injections, arthroscopy), therapeutic apheresis, and interventions related to pharmacological or surgical interventions.

  • Outcomes: For the purpose of this overview, the primary outcome measures were pain, stiffness, and function. The concept of “function” is based on the International Classification of Functioning, Disability and Health (ICF)11 definition, where “function” is an umbrella term for body functions, body structures, activities, and participation.1,9

We searched the Cochrane Library (Cochrane Database of Systematic Reviews and DARE), MEDLINE, EMBASE, PEDro, PsychINFO, and CINAHL from 2000 up to week 40 of 2008 for “hand osteoarthritis/arthrosis or OA.” A broad computerized search strategy was developed (Appendix 1). Reference lists from retrieved reviews were examined.

Retrieved hits were assessed by 2 of the authors (R.H.M., K.B.H.), who screened the titles and abstracts to identify relevant studies. If doubt occurred, one of the other authors was consulted. The full text of potential relevant articles was read by 2 authors (R.H.M., K.B.H.).

Assessment of Methodological Quality

Two authors (R.H.M., K.B.H.) independently assessed the methodological quality of the reviews. Disagreement was resolved by discussion. Eleven criteria were rated as “met,” “unclear/partly met,” or “not met” according to a criteria list from the Measurement Tool to Assess Systematic Reviews (AMSTAR) for assessing quality of evidence for each review. AMSTAR is a reliable and valid measurement tool for assessing systematic reviews based on assessments of quality of primary studies, design of primary studies, consistency, and directness, with overall scores ranging from 0 to 10 (out of a maximum of 11 criteria)12,13 (Appendix 2).

Data Extraction and Synthesis

Data on effectiveness were extracted from the identified high-quality reviews by 2 of the authors (R.H.M., K.B.H.). The following criteria were applied when data on effects were extracted:

  • Adequate quantitative pooling of data in reviews was regarded as more valid than a qualitative data synthesis approach.

  • If no direct comparisons between treatments were undertaken or no quantitative pooling of data was performed, the results were reported as “no quantitative pooling,” and the statement by the authors was reported.

  • When we found that the results were reported inconsistently in different sections of one review or inconsistently between 2 different reviews, the effects were extracted from the referenced primary studies.


Selection Procedure

The literature search identified 173 reviews on hand OA. One hundred sixty-five articles were clearly not relevant based on information from the title and abstract. The full text of 8 articles was retrieved and assessed, and 3 articles were excluded for various reasons (Figure). Based on the assessment of the methodological quality of the remaining 5 systematic reviews, one review14 was excluded because it met none of the AMSTAR quality criteria for systematic reviews. Thus, 4 systematic reviews7,1517 were included and served as the basis of this umbrella review (Tab. 1). One of these reports is not strictly a systematic review, but offers recommendations for treatment.7 As it includes all of the core elements of a systematic review, we decided to include it in the present overview. The characteristics and methodological assessment of the 4 included reviews are presented in Table 1.



Table 1.
Table 1.

Characteristics of Included Reviewsa



Generally, few reviews provided results by quantitative pooling. For one of the included reviews, results in terms of effects were not extracted because outcomes for the comparisons (ie, pain or function) were not specified.16


As shown in Table 2, topical capsaicin, the active principal of hot chili pepper, was more effective than a placebo in reducing pain. Zhang et al7 reported that the number needed to treat to obtain moderate to excellent (more than 50%) pain relief or symptomatic improvement was 3 (95% confidence interval=2–5).

Table 2.
Table 2.

Results Compiled From 4 Systematic Reviews on Nonpharmacological and Nonsurgical Intervention for Hand Osteoarthritis (OA)a


For splinting of the thumb, there were conflicting results between 2 of the reviews on both the quantification of effect sizes and the direction of effect.7,17 Zhang et al presented a quantitative pooling of 2 RCTs18,19 and concluded that there was “more pain relief from the full splint compared to the half splint (ES [effect size]=0.64 [0.02–1.26]).”7 Egan and Brousseau did not provide any quantitative pooling, but reported that “there was fair evidence for the effectiveness of splinting to relieve pain and improve function.”17 However, they acknowledged that “this evidence came exclusively from pretest-posttest types of studies or from the pretest-posttest phase of RCTs.”17 Based on 3 RCTs comparing different kinds of splinting,1820 Egan and Brousseau further stated, “There was no clear evidence of the superiority of one type of splint over another for pain relief, comfort or function.”17

Egan and Brousseau included one trial (Buurke et al20) that was not included by Zhang et al.7 However, this trial included only 10 women who wore 3 different splints in random order, and it was not possible to calculate effect sizes “due to unreliability in data.”17 Weiss et al19 (N=26) compared a custom-made short opponens splint (which crosses only the carpometacarpal [CMC] joint) with a large opponens splint (which crosses the CMC joint, the wrist, and the metacarpophalangeal joint) and found that both splints reduced CMC joint pain significantly (pretest-posttest), but there was no significant difference between the splints in their effect on thumb pain. Although Egan and Brousseau presented no data for statistical recalculation, they concluded that “our calculations based on their graphs demonstrated the superiority of the short opponens splint.”17 Weiss et al18 (N=25) compared a custom-made short opponens thermoplastic splint (which crosses only the CMC joint) with a prefabricated neoprene splint (which crosses both the CMC joint and the metacarpophalangeal joint). They found that thumb pain was significantly less (P=.019) when wearing the neoprene splint compared with the thermoplastic splint. Considering that these trials had relatively small sample sizes, with unclear and somewhat conflicting effect estimates, it may be reasonable to conclude that there is limited evidence that splints for the CMC joint in people with OA have a pain-reducing effect, but there is not enough evidence to give any recommendations regarding design or material.

For the effect of thermal vapor treatment at 44°C versus topical ibuprofen on pain (1 RCT, N=116), no quantitative pooling was presented. The authors’ conclusion was presented as: “No significant difference.”15


For the comparison between exercise and education versus OA information only, the effects based on 1 RCT (N=40) was in favor of exercise and education. The number needed to treat to obtain moderate to excellent (more than 50%) improvement in “patient global function” on a visual analog scale was 2 (95% confidence interval=1–6).7

For the effect of thermal vapor treatment at 44°C versus topical ibuprofen on function (1 RCT, N=116), no quantitative pooling was presented. The authors’ conclusion was presented as: “Significant vs control.”15 The authors further stated that “Berthollet was superior over topical ibuprofen at treatment completion.”15(p145)

Other outcomes.

On the question of whether patients’ desire for surgery changed after splinting, the comparisons were splinting versus no treatment.17 This review was based on a single RCT (N=33), and no quantitative pooling was presented. The authors’ conclusion was: “Approximately one third of each group desired surgery, indicating that splinting did not have an effect on this outcome.”17(p74)


Based on an overview of systematic reviews, we found that there is some evidence for the pain-relieving effect of topical capsaicin compared with a placebo and favorable functional outcomes for exercise and education compared with OA information alone. There also is limited evidence that splinting of the thumb CMC joint reduces pain. However, the most striking finding of the present overview was the paucity of available systematic reviews. Thus, there currently is a very limited body of evidence for the effects of nonpharmacological and nonsurgical interventions for hand OA.

The present overview may have several limitations. A major limitation applies to summarizing evidence based on systematic reviews, as new, relevant primary studies may have been published but were not captured in the included systematic reviews. The review of Zhang et al7 is the most recent of the included reviews and addresses all nonpharmacological treatments. Zhang et al included studies published up to January 2006 and could identify only a few relevant primary studies (RCTs and controlled clinical trials). For example, they identified only 1 study on education, 2 studies on exercises and yoga, 1 study on transcutaneous electrical nerve stimulation, 1 study on laser treatment, and 3 studies on splints and gloves. It is unlikely that a substantial number of new studies that could alter our findings have been published since then. Other limitations of overviews are that important information from primary studies is not reported and that overviews are too broad to be useful for clinicians. We would argue that overviews are important tools to guide directions in choice of interventions, but they might not be useful for deciding how interventions specifically should be carried out.

Furthermore, the results of an umbrella review could be limited by the presence of publication bias at the level of primary studies. Although “publication bias” was assessed, according to AMSTAR, we cannot exclude that publication bias at the level of primary studies might have biased the findings in the present overview. In an umbrella review, it is very important to include all systematic reviews that meet the inclusion criteria. A broad computerized search strategy was performed (Appendix 1) in addition to hand searches, but we still might have missed eligible systematic reviews. We excluded reviews providing anecdotal evidence for clinically relevant (and potentially effective) interventions, but will argue that the main aims of such an overview are to provide clinicians, patients, and policy makers with unbiased information about the effectiveness of nonpharmacological interventions for hand OA and to provide the research community with unbiased information about research gaps on this topic.

One of the reviews identified through the electronic searches was excluded based on the assessment of the methodological quality.14 It was published in a journal supplement and originally presented as a critical review, not a systematic review. The primary studies included were all covered by the other 4 reviews included in the present overview. Thus, including this review would not have significantly influenced the results. One of the reviews7 was published as a treatment recommendation, but we included it because it contained a systematic review of available treatments for hand OA. The other components of the article leading to treatment recommendations were not considered in this study.

All reviews, regardless of whether they are systematic reviews or umbrella reviews, rely on the availability of high-quality studies of the effects of different interventions. Although the paucity of systematic reviews for the purpose of this study was striking, it also indicates a lack of high-quality primary studies for the treatment of people with hand OA—which in itself is an incentive for systematic reviews. Thus, in this overview, in addition to summarizing the evidence on treatment of people with hand OA that does not include drugs or surgery, we identify a demand for primary studies on exercise, education, and splints, which could alleviate the burden of the disease to patients.


Based on an overview of available systematic reviews, there currently is insufficient high-quality evidence regarding nonpharmacological and nonsurgical interventions for hand OA. Considering the limited literature in this area and the prevalence and impact of the disease, more primary studies and updated systematic reviews are warranted.

Download the file : 1363.full.pdf (1.1 MB) Find it online