Systematic review of non-surgical therapies for osteoarthritis of the hand: an update.

Authors: Mahendira D (1) , Towheed T (1)
(1) Department of Medicine, Queen's University, Kingston
Source: Osteoarthritis Cartilage. 2009 Oct;17(10):1263-8
DOI: 10.1016/j.joca.2009.04.006 Publication date: 2009 Oct E-Publication date: April 17, 2009 Availability: full text Copyright: © 2009 Osteoarthritis Research Society International. Published by Elsevier Inc.
Language: English Countries: Not specified Location: Not specified Correspondence address: Tanveer E. Towheed,
Department of Medicine, Queen's University, Room 2066, Etherington Hall, Kingston, Ontario, Canada, K7L 3N6.
Tel: 1-613-533-6896; Fax: 1-613-533-2189.
Email :


Article abstract


To update our earlier systematic review which evaluated all published randomized controlled trials (RCTs) evaluating pharmacological and non-pharmacological therapies in patients with hand osteoarthritis (OA). Surgical therapies were not evaluated.


RCTs published between August 2004 and February 2008 were added to the original systematic review.


A total of 44 RCTs evaluating various pharmacological and non-pharmacological therapies in hand OA were analyzed in this update. Generally, these RCTs were of low quality. RCTs were weakened by a lack of consistent case definition and by a lack of standardized outcome assessments. The methods used for randomization, blinding, and allocation concealment were rarely described. The number and location of symptomatic hand joints per treatment group at baseline was usually not stated. The number and location of evaluated hand joints at the end of the study was also usually not stated. A meta-analysis could not be performed since most of the treatments studied did not have more than one identical comparison to allow pooling of the data.


It is apparent that hand OA is a more complex area in which to study the efficacy of therapies when compared to hip and knee OA. The recently published OARSI Consensus Recommendations will improve the design and conduct of future RCTs in hand OA.

Article content


Therapy for hand osteoarthritis (OA) has received relatively little attention when compared to OA of the hip and knee. The objective of this paper is to update our previous version of the systematic review of therapies for OA of the hand by adding randomized controlled trials (RCTs) published between August 2004 and February 20081.


Criteria for inclusion and exclusion

The inclusion and exclusion criteria were identical to those used in the original version of the systematic review. RCTs evaluating surgical therapies were not included. We included RCTs evaluating OA at multiple sites only if efficacy data were presented separately for the hand.


Search strategy and study identification

The following electronic data sources were searched for this updated version of the systematic review: (1) MEDLINE (1966 to February week 4, 2008), (2) PREMEDLINE (February week 4, 2008), (3) EMBASE (1980 to February week 4, 2008), (4) AMED (1985 to February week 4, 2008), (5) CINAHL, Allied Health (1982 to February week 4, 2008), (6) EBM Reviews, including the Cochrane Database of Systematic Review (CDSR), the Database of Abstracts of Reviews of Effectiveness (DARE), ACP Journal Club and the Central Cochrane Database (1980 to February week 4, 2008). Reference lists of all retrieved articles were also manually searched. The flow chart summarizing study identification and retrieval is shown in Fig. 1. The search strategy used in MEDLINE was identical to that used in the earlier version of this review.

Fig. 1

Flow chart of search strategy.



A data abstraction form was used to extract information pertaining to trial demographics, methodology, quality, and outcomes2, 3. Study quality was evaluated by using Jadad's scoring checklist4. The final score ranges from 0 to 5, with a higher score reflecting higher methodological quality. Data abstraction was conducted by the two authors, independently. Results were cross-checked for reliability and differences were resolved by reaching consensus. Allocation concealment was specifically evaluated for each RCT. A formal meta-analysis will be performed, if feasible.




Overview of study demographics

A total of 44 RCTs were analyzed in this systematic review5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47 (Table I, Table II, Table III). Two RCTs were published in the 1970s, five were published in the 1980s, 14 were published in the 1990s and 23 were published in the year 2000 or beyond. Thirty-six RCTs were available as English full paper reports, four were non-English reports with English abstracts, and four were only available as English abstracts.

Table IPublished RCTs in OA of the hand (N=44)
Study (authors and year) Groups N randomized N completed Design Duration (weeks) Overall efficacy
Thorpe, 19708 Fiorinal vs FIPA vs placebo 10 9 Cross-over 6 (Fiorinal=FIPA)>placebo
Swezey et al., 197928 Pressure glove vs control glove vs no glove 5 5 Cross-over 6 Equal
Seiler, 198313 Meclomen vs placebo 41 22 Parallel 4 Meclomen>placebo
Talke et al., 198532 Topical etofenamate vs oral indomethacin NA NA Parallel 3 Equal
Basford et al., 198723 Helium neon laser vs placebo 81 81 Parallel 3 Equal
Caruso et al., 198733 SAMe† vs naproxen vs placebo 51 NA Parallel 4 Equal
Pastinen et al., 19887 Glycosaminoglycan polysulfate (GAGPS) IA‡ vs placebo 30 29 Parallel 52 GAGPS>placebo
McCarthy and McCarty, 199226 Capsaicin vs placebo 14 14 Parallel 4 Capsaicin>placebo
Dreiser et al., 199315 Ibuprofen vs placebo 60 54 Parallel 2 Ibuprofen>placebo
Verbruggen and Veys, 199319 GAGPS (IM)§ vs placebo 92 68 Parallel 260 GAGPS>placebo
Flynn et al., 199422 Folate vs folate+B12 vs placebo 30 26 Cross-over 24 Folate+B12>(placebo=folate)
Garfinkel et al., 199425 Yoga vs no therapy 26 25 Parallel 10 Yoga>no therapy
Rothacker et al., 19946 Trolamine salicylate vs placebo 50 49 Cross-over NA Trolamine>placebo
Schnitzer et al., 199417 Capsaicin vs placebo 59 48 Parallel 9 Capsaicin>placebo
Dougados and Nguyen, 199521 Topical niflumic acid vs placebo 186 186 Parallel 1 Equal
Thiesce and Dougados, 199520 Topical diclofenac vs placebo 20 20 Cross-over 1.5 Equal
NA=not available.
Table IIPublished RCTs in OA of the hand (N=44)
Study (authors and year) Groups N randomized N completed Design Duration (weeks) Overall efficacy
Renklitepe et al., 199531 Tens electrode glove vs carbon electrode 36 NA Parallel 0.7 Glove electrode>carbon electrode
Punzi et al., 199634 Hydroxychloroquine (HQ) vs NSAID/analgesics 15 15 Parallel 52 HQ>NSAID+analgesic
Graber-Duvernay et al., 199729 Berthollet spa vs topical ibuprofen 116 107 Parallel 24 Spa>ibuprofen
Rothacker et al., 199814 Trolamine salicylate vs placebo 86 81 Parallel 0.01 Trolamine>placebo
Buurke et al., 199912 Uriel splint vs sporlastic splint vs gibortho splint 10 10 Cross-over 12 Uriel splint>others
Reeves and Hassanein, 200011 Dextrose prolotherapy (DP) vs placebo 27 25 Parallel 24 DP>placebo
Weiss et al., 20009 Short splint vs long splint vs no splint 26 26 Cross-over 2 Short splint>long splint>no splint
Randall et al., 200016 Stinging nettle leaf vs placebo 27 24 Cross-over 12 Stinging nettle leaf>placebo
Berggren et al., 200110 OT vs OT+textile splint vs OT+leather splint 33 33 Parallel 28 All groups had less hand surgery
Zacher et al., 200130 Topical diclofenac vs oral ibuprofen 321 NA Parallel 3 Equal
Stamm et al., 200227 Joint protection and exercise (JPE) vs info only 40 40 Parallel 12 JPE>info only
Verbruggen et al., 20025 Chondroitin polysulfate (CPS) vs placebo 130 92 Parallel 156 CPS>placebo
Verbruggen et al., 20025 Chondroitin sulfate (CS) vs placebo 92 73 Parallel 156 CS>placebo
Lisse et al., 200324 Rofecoxib vs naproxen 910 NA Parallel 12 Equal
Rovetta et al., 200418 CS and naproxen vs naproxen alone 24 24 Parallel 104 CS+naproxen>naproxen alone
NA=not available.
Table IIIPublished RCTs in OA of the hand (N=44)
Study (authors and year) Groups N randomized N completed Design Duration (weeks) Overall efficacy
Grifka et al., 200435 Lumiracoxib vs placebo 594 559 Parallel 4 Lumiracoxib>placebo
Lefler et al., 200436 Strength training exercises vs control 19 18 Parallel 6 Strength training>control
Meenagh et al., 200437 IA corticosteroid vs placebo 40 35 Parallel 24 Equal
Weiss et al., 200441 Neoprene splint (PFN) vs custom-made splint (CMS) 25 25 Cross-over 2 PFN>CMS
Brosseau et al., 200538 Low level laser therapy vs placebo 88 86 Parallel 6 Equal
Stahl et al., 200539 IA corticosteroid vs IA hyaluronate 52 52 Parallel 24 Equal
Wajon et al., 200540 TSS+exercise vs SOS†+exercise 40 34 Parallel 6 Equal
Fuchs et al., 200642 IA hyaluronate vs IA steroid 56 51 Parallel 26 Equal
Michalsen et al., 200643 Leeches vs topical diclofenac 32 32 Parallel 8.6 Leeches>diclofenac
Stange-Rezende et al., 200644 Infrared radiation (IRR) vs control 45 35 Cross-over 8 IRR>control
Roux et al., 200745 IA hyaluronate (once vs twice vs thrice) 42 37 Parallel 12 Equal
Widrig et al., 200746 Topical ibuprofen gel vs arnica gel 204 174 Parallel 3 Equal
Heyworth et al., 200847 IA hylan vs IA corticosteroid vs placebo 60 60 Parallel 26 Equal
TSS=thumb strap splint.
SOS=short opponens splint.

Thirty-two reports evaluating therapies in hand OA were excluded from this review since they did not meet one or more of the stated inclusion criteria of this systematic review48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 76, 77, 78, 79.

Symptom modifying therapy was evaluated in 39 RCTs, whereas a structural modifying therapy was evaluated in three RCTs5, 19. Two RCTs evaluated both a symptom modifying as well as a structural modifying therapy11, 18. A parallel, independent group study design was used in 35 RCTs and a cross-over design was used in nine RCTs. The median number of subjects randomized per study was 41, with a range of 5–910. The median number of subjects completing the trials was 35, with a range of 5–559.

The median duration of the RCTs was 8.6 weeks, with a range of 2 h–260 weeks. Eighty-two percent of the subjects randomized were female. The mean age of randomized subjects in the RCTs was 62.3 years, with a range of 53–82 years. Subjects had a mean duration of OA hand symptoms of 6.1 years, with a range of 2–10.2 years.

Five RCTs had an open follow-up period after study discontinuation. Twenty-seven of the 44 RCTs (61%) had a placebo group/arm. Seven RCTs were multi-centre. The country of origin was heterogeneous with the US having the greatest number of RCTs (14), followed by France (5), Germany (5), Austria (3), Belgium (3), and Italy (3).

Nine RCTs had an oral non-steroidal anti-inflammatory drug (NSAID) treatment group/arm. The individual NSAIDs evaluated were: ibuprofen, naproxen, meclofenamate, rofecoxib, lumiracoxib and indomethacin. Seven RCTs had a topical NSAID treatment group/arm. The individual topical NSAID preparations evaluated were: topical ibuprofen, topical etofenamate, topical diclofenac, and topical niflumic acid. Two RCTs had a topical ASA group/arm. Both studies evaluated topical trolamine salicylate in comparison to a placebo. Four studies evaluated intra-articular steroid use, while four studies examined intra-articular hyaluronan. Seven RCTs evaluated occupational therapy interventions. These included the following: joint protection exercises, splints of various kinds and technical accessories. Other active agents tested included: capsaicin cream (2), glycosaminoglycan polysulfate (GAGPS) (2) and chondroitin sulfate (2). Several RCTs evaluated unconventional OA therapies, such as: laser therapy (2), folate and cobalamin vitamin therapy (1), leeches (1), tiled stove (1), TENs (1), yoga (1), dextrose prolotherapy (1), pressure gradient gloves (1), spa therapy (1), stinging nettle leaf (1), fiorinal of (1) and hydroxychloroquine (1).


Features specific to hand OA trials

No consistent definition of hand OA was used in these RCTs. Most trials (N= 32) did not explicitly distinguish between primary (idiopathic) and secondary OA. Ten RCTs evaluated exclusively subjects with primary hand OA. One RCT enrolled subjects with both primary and secondary hand OA. Erosive hand OA was studied exclusively in two RCTs18, 34. Only 11 RCTs used a validated hand OA classification scheme for study entry. These 11 studies used the ACR classification criteria80. Most often, hand OA was defined by the authors themselves with descriptions that lacked sufficient detail and precision (N=23 RCTs). Radiographs were taken at baseline in 24 RCTs. However, detailed X-ray criteria were explicitly stated in only 16 of these 24 RCTs.

The distribution of affected hand OA joints was quite variable and was also inconsistently presented in the trial reports. For example, 11 RCTs did not specify which joints were being evaluated in the hand. Twelve RCTs evaluated exclusively subjects with first carpal metacarpal joint (CMC) OA. Eleven RCTs evaluated subjects with interphalangeal (proximal interphalangeal joint (PIP) and/or distal interphalangeal joint (DIP)) OA, but not first CMC OA. Nine RCTs evaluated all three joint areas (PIP, DIP and first CMC).

Only 15 RCTs presented detailed information pertaining to the number and location of symptomatic hand joints per treatment group at baseline. This information is important to include as a clinical descriptive feature of how much patients are affected by hand OA. Only 19 RCTs specified the number and location of evaluated hand joints at the end of the study. Only 15 RCTs specified a priori a hand joint site for efficacy analysis.

Eight RCTs did not specify whether one hand or both hands were evaluated. Seventeen RCTs evaluated one hand only, whereas 18 RCTs evaluated both hands. Fifteen RCTs specified a minimum entry criterion for symptoms and/or objective findings, including specific X-ray criteria at baseline.

There was a lack of standardization of outcomes across the 44 RCTs. Moreover, some outcome variables used in these RCTs have not been validated in OA trials. Pain, function, and patient global assessments were evaluated in 38, 35 and 19 RCTs, respectively. Physician global assessments and health-related quality of life were evaluated in nine and three RCTs, respectively. Examples of outcome variables used in these trials that have not been validated in OA trials include joint swelling, joint tenderness, need for OA related surgery, analgesic usage, sleep quality and range of motion. A standardized, validated questionnaire (generic and/or disease specific) was used for outcome assessment in 18 RCTs. This included the disease specific AUSCAN questionnaire81 and Dreiser's Functional Index82.


Features of trial quality

Pre-randomization inclusion criteria were clearly specified in 36 RCTs. Pre-randomization exclusion criteria were clearly specified in 30 RCTs. Subjects were blinded in 25 RCTs and investigators were blinded in 29 RCTs. Fifteen of the 44 RCTs were associated with a pharmaceutical company or manufacturer. Six RCTs excluded subjects for protocol violation. Eleven RCTs excluded subjects for adverse effects. Thirty-three RCTs did not specify as to whether subjects had prior exposure to the test agents. Fifteen RCTs controlled for supplementary analgesic usage or co-intervention during the trial duration. Ten RCTs presented sample size calculations. Nine RCTs provided a clearly stated rationale for the chosen dosage of at least one active agent. Eleven RCTs described the method of randomization. Fourteen RCTs described the method of blinding. Seventeen RCTs defined a priori a main outcome variable. The success of the blinding was only evaluated at the end of the study in one of the included RCTs. Eighteen RCTs presented sufficiently detailed baseline data allowing the reader to ensure that the groups were comparable at baseline. Twenty-four RCTs used appropriate statistical analyses. Examples of inappropriate statistical analyses included: (1) using a parametric statistical test for non-parametric data, (2) stating that a marginally insignificant statistical test was still statistically “significant”, (3) using a paired statistical test for independent groups, (4) using multiple comparisons without employing any statistical correction. For example, one RCT evaluated several outcome variables, at separate time points, for the right hand, left hand, both hands, and the dominant hand. Twenty-one RCTs either had no withdrawals or used an intention-to-treat analysis. Only four RCTs adequately described the method used to ensure allocation concealment83.


Methodological quality based on Jadad's scores

The median Jadad score for the entire group of RCTs was 3, with a range of 0–5. Twenty-five RCTs received a score of 1 for being double-blinded. Twenty-nine RCTs received a score of 1 for providing an adequate description of withdrawals and drop-outs. All RCTs were randomized as per the inclusion criteria of the systematic review and therefore all received a score of 1 for this criterion. The median Jadad score for the 13 RCTs that were added to this version of the systematic review was only 2. Thus, it does not appear that the quality of the more recent hand OA trials has improved.



A formal meta-analysis could not be performed as most of the therapies studied did not have more than one identical comparison to allow pooling of the data. Also, the studies were too clinically diverse making a meta-analysis inappropriate.


Summary of results of therapies

Due to the inherent methodological limitations in these studies, it is difficult to offer any reliable practical recommendations for the choice of appropriate therapy in subjects with clinically significant hand OA (Table I, Table II, Table III). However, based on these 44 analyzed RCTs, one can conclude that there is at least some evidence from a published RCT for the efficacy for the following therapies: trolamine salicylate, GAGPS, fiornal and FIPA, splints for first CMC OA, occupational therapy, dextrose prolotherapy, oral NSAIDs, stinging nettle leaf, topical capsaicin, vitamin B12 with folate, yoga, leeches, strength training exercise and spa therapy. At least some evidence also exists from a published RCT for structural efficacy in hand OA for the following therapies: chondroitin sulfate, chondroitin polysulfate, and GAGPS.



Results of this systematic review allow for a number of general conclusions. First, there are few published RCTs evaluating the wide range of therapies available for hand OA. However, 13 RCTs were published from August 2004 to February 2008, which is a relative increase from previous years. Second, RCTs in hand OA are weak methodologically. The most important issues relate to deficiencies in allocation concealment, inadequate description of the methods of randomization and blinding, failure to use intention-to-treat analyses, inappropriate statistical analyses, and failure to provide sample size calculations. Third, a number of more specific methodological limitations related to hand OA are also apparent. The number and location of symptomatic joints per treatment group at baseline was usually not stated. In addition, the number and location of evaluated joints per treatment group at the end of study was also usually not stated. RCTs were often lacking important details regarding whether one or both hands were evaluated, whether the first CMC joint and/or IP joints were evaluated, and the selection a priori of both a target joint site (first CMC vs IP joints) and a principal outcome measure for efficacy evaluation. These are some of the issues that were addressed by the OARSI task force in the recently published Consensus Guidelines for the Design and Conduct of trials in subjects with hand OA83.

Download the file : PIIS1063458409001058.pdf (146.7 KB) Find it online