Thirty RCTs involving a total of 31 comparisons were included (n at least 2,648). There were 7 crossover comparisons from crossover RCTs and 24 comparisons from parallel-group RCTs. The sample size, where reported, ranged from 5 to 910 at enrolment.
Study quality was generally low (median Jadad score 3, range: 0 to 5). Methodological limitations included the use of inappropriate statistical analysis (16 RCTs) and lack of the following: patient blinding (11 RCTs), investigator blinding (13 RCTs), control for cointerventions (20 RCTs), sample size calculations (26 RCTs), description of randomisation method (25 RCTs), description of method of blinding (22 RCTs), definition of primary outcome (22 RCTs), data to allow comparison of baseline characteristics between treatment groups (20 RCTs), and description of adequate method of allocation concealment. Other problems included inconsistent definitions of OA of the hand, a lack of standardised methods to assess outcomes, and insufficient reporting of the number and location of joints at study entry and at follow-up.
There was some evidence of efficacy from an RCT for the following therapies: trolamine salicylate, glycosaminoglycan polyphosphate, fiornal and FIPA, splints for first carpal metacarpal OA, occupational therapy, dextrose prolotherapy, oral non-steroidal anti-inflammatories, stinging nettle leaf, topical capsaicin, vitamin B12 with folate, yoga and spa therapy.
There was also some evidence from an RCT for structural efficacy of glycosaminoglycan polyphosphate, chondroitin sulphate and chondroitin polysulphate.