Background: Hand eczema is a major cause of morbidity and lost earnings. Many interventions ranging from topical steroids to oral ciclosporin are used, but their evidence base and the best methods to assess their efficacy are uncertain.
Objectives: As part of a long-term project to improve standards of design and reporting in hand eczema trials, we sought to describe the prevalent study designs and comment on the quality of reporting of such studies. METHODS AND DATA SOURCES: Electronic databases (Cochrane, Medline, Embase, Pascal, Jicst-Eplus, Amed) were searched from January 1977 to April 2003 using all possible variants of the terms hand and eczema/dermatitis. In addition, four general medical and 17 specialist dermatology journals were hand-searched by pairs of researchers for all possible therapeutic studies.
Study selection: Studies were eligible for inclusion if they dealt with hand eczema as diagnosed by a physician irrespective of the aetiology, and if they described the results of a study of a therapeutic intervention in humans. Single case reports and reviews were excluded, but case series and nonrandomized studies were considered alongside randomized studies. Data selection For each study, two researchers independently assessed the type of study, outcome measures, enrolment criteria, randomization, masking of interventions and how losses to follow-up were dealt with.
Main outcome measures: Proportion of studies according to type of intervention and study type. Proportion of randomized controlled trials (RCTs) that adequately reported eligibility criteria, randomization generation and concealment, masking and intention-to-treat analysis.
Results: A total of 90 studies reported in 87 papers dealt with 11 different classes of interventions. Around 80% of the studies dealt with just four interventions: ultraviolet light, topical steroids, radiation and systemic immunosuppressives. Of the 90 studies, 44 were case series, 15 were nonrandomized controlled trials, and the remaining 31 were RCTs. Of the 31 RCTs, 16 were parallel (one with cross-over design) and 15 self-controlled. Only 11 of the RCTs adequately reported eligibility criteria. The randomization method was described in 10, and there was adequate concealment of allocation in eight. Masking the treatment allocation from both the study assessors and patients was done in 11 RCTs, and intention-to-treat analysis was reported in four. Only 13 RCTs were 4 months or longer in duration. No study reported a rationale for the sample size, and in only one study had the outcome variable been validated.
Conclusions: Most 'trials' in hand eczema are not RCTs. Internally controlled (left/right) studies were common. Based on the poor overall quality of reporting, most RCTs of hand eczema trials are not adequate to guide clinical practice. Future trials of hand eczema should be randomized, using a parallel group or self-controlled design. Research is needed to develop validated and clinically relevant outcome measures. Most of the remaining issues relating to poor quality of existing evidence can be relatively easily dealt with by following the CONSORT guidelines.