Ruzicka 2008.

Methods	Parallel‐group, randomised controlled trial including 1 placebo group and 2 treatment groups given different doses of the same (oral) retinoid This study was carried out in a secondary care setting This was a multi‐centre study at 111 clinics in Europe and Canada
Participants	1032 participants (582 male, 450 female) with severe chronic hand dermatitis of at least 6 months' duration and refractory to standard therapy. All types of hand dermatitis Dropouts: 273 Inclusion criteria of the trial Severe chronic hand eczema refractory to standard therapy 18 to 75 years of age Exclusion criteria of the trial Well defined Study population Gender: placebo group 84 female, 121 male; alitretinoin 10 mg group 180 female, 238 male; alitretinoin 30 mg group 186 female, 223 male Age: placebo group mean 48 years, SD 12 years; alitretinoin 10 mg group mean 47 years, SD 13 years; alitretinoin 30 mg group mean 48 years, SD 13 years
Interventions	Intervention • Oral alitretinoin 10 mg once daily for 12 or 24 weeks (depending on moment of response according to the PGA) in 319/418 participants • Oral alitretinoin 30 mg/d in 303/409 participants for 12 or 24 weeks Control intervention • Placebo capsules in 137/205 participants for 12 or 24 weeks Standard emollient in all treatment groups All participants were followed up for 4 weeks, and responders were observed for relapses for 24 weeks after end of treatment Duration Up to 48 weeks (12 to 24 weeks of active treatment, up to 24 weeks of follow‐up)
Outcomes	Primary outcome of the trial Responders according to physician global assessment of overall severity, whereby physician global assessment is categorised as clear, almost clear, mild, moderate, or severe. Responders are defined as clear or almost clear at week 12 or at last evaluation Other outcomes Time to response Partial response (PGA assessment of clear, almost clear, or mild) Observer‐rated modified total lesion symptom score: sum of scores (0 = absent, 1 = mild, 2 = moderate, 3 = severe) for erythema, oedema, vesicles, desquamation, hyperkeratosis, fissures, pruritus/pain Participant‐rated global assessment: clearing or almost clearing (> 90% clearing of signs and symptoms compared with baseline), marked improvement (> 75%), moderate improvement (> 50%), mild improvement (> 25%), no change, worsening Time to relapse Observer‐rated extent of disease: total percentage involvement of palm and dorsum of both hands Adverse events
Notes	No other active treatment as comparator. Analysis of efficacy based on intention‐to‐treat principle. Study included a safety assessment by careful medical and laboratory monitoring. More males were enrolled because of exclusion of women of child‐bearing potential Declarations of interest: some study authors were employees, received grants, or had received consultancy fees from Basilea Pharmaceutica Funding: see above item Sample size rationale: provided
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "eligible patients were randomized to treatment by centre, in blocks of 5 without stratification, by the use of computer‐generated randomization codes provided by the study sponsor (Basilea Pharmaceutica)..." Comment: computer‐generated randomisation codes
Allocation concealment (selection bias)	Low risk	Quote: "eligible patients were randomized to treatment by centre, in blocks of 5 without stratification, by the use of computer‐generated randomization codes provided by the study sponsor (Basilea Pharmaceutica) and incorporated into double‐blind coded drug packaging. Placebo, active drug and packaging were indistinguishable. Investigators allocated consecutively numbered packages of medication to patients in their order of enrolment" Comment: codes provided by study sponsor. Sequentially numbered packages of different treatment modalities of identical appearance were used
Blinding of participants and personnel (performance bias) All outcomes	Low risk	Quote: "... incorporated into double‐blind coded drug packaging. Placebo, active drug and packaging were indistinguishable" Comment: double‐blinded; the identical looking packages were provided by the sponsor and by site staff, and participants were unaware of the treatment
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Observers were blinded during the study because the identical looking package of study drugs was provided by a third party. One might argue that the observer could have guessed the treatment group due to headache and dry mucosa; however these were also seen in the control group and therefore were not conclusive. The trial was designed in such a way as to minimise risk of bias; we agree that this could not have been done in a better way
Incomplete outcome data (attrition bias) All outcomes	Low risk	Quote: "all efficacy evaluations were based on the intent‐to‐treat population. All randomized patients were included in this population, and were analysed according to their randomization with last observation carried forward (LOCF) in cases of missing data" Comment: intention‐to‐treat analysis
Selective reporting (reporting bias)	Unclear risk	Trial registration on clinicaltrials.gov (NCT00124475). No differences in primary outcomes; however small discrepancies in other outcomes between trial registration and article
Other bias	Low risk	Baseline comparisons: no significant differences between groups in demographic or disease characteristics Diagnostic certainty: yes The study was completed