Mabin 1995.

Methods	This was a randomised controlled trial (single centre, UK). D = parallel AC = clear ‐ randomisation schedule supervised by the Director of Pharmacy RS = random number tables using blocks of 4 B = outcome assessor
Participants	Inclusion criteria of the trial 48 children (T1: 24; T2: 24) older than 12 months (age range 2 to 15 years) who had not received oral corticosteroids in the last 2 months. 41 were evaluable (T1:19, T2: 22) Setting = outpatients  SAE = moderate or severe Dig = Hanifin and Rajka Co‐T = all participants were receiving conventional topical treatment with emollients and corticosteroids of milk or moderate potency Exclusion criteria of the trial Children under 12 months Those who had received oral corticosteroids in the preceding 2 months
Interventions	T1: pyridoxine hydrochloride (n = 24) T2: placebo tablet (n = 24) T1 vs T2 ‐ 1 tablet (50 mg) per day for 4 weeks
Outcomes	FU at baseline and after 4 weeks of treatment Parents recorded in a diary the amount of nocturnal sleep disturbances due to scratching, and the amount of daytime scratching on a numeric scale from 0 (none) to 10 (the worst you could imagine) The percentage surface area of the body affected by dermatitis was determined using charts that divided the body into 32 separate zones. Skin severity for each of the 32 zones assessed by extent of area affected. The total of all 32 zones summated to give total extent of skin involvement Overall degree of erythema graded on an arbitrary scale: 0 (none) to 5 (severe) After 4 weeks parents recorded if eczema was 'better', 'worse', or 'no change'
Notes	A per‐protocol analysis was undertaken. 7 participants were lost to FU ‐ 3 from the treatment group, 2 from the placebo group (3 because of eczema flare up due to bacterial infection requiring antibiotics, and 2 due to exacerbations of coexistent asthma requiring oral prednisolone treatment), 1 stopped taking the tablets suspecting food allergy, and 1 failed to return for the 4‐week assessment. Randomisation led to differences in sex distribution, with 79% men in the treatment group and 68% women in the placebo group. The scoring of the degree of erythema was semi‐objective. The dosage was the same regardless of the age or size of the child. There was no table of results.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Random number tables using blocks of 4 were used.
Allocation concealment (selection bias)	Low risk	Randomisation schedule were supervised by the Director of Pharmacy.
Blinding (performance bias and detection bias) All outcomes	Low risk	The outcome assessor was blinded.
Incomplete outcome data (attrition bias) All outcomes	Low risk	7 participants were lost to FU ‐ 3 from the treatment group, 2 from the placebo group (3 because of eczema flare up due to bacterial infection requiring antibiotics, and 2 due to exacerbations of coexistent asthma requiring oral prednisolone treatment), 1 stopped taking the tablets suspecting food allergy, and 1 failed to return for the 4‐week assessment.
Certainty of AD	Low risk	This was measured using criteria by Hanifin and Rajka.
Baseline comparability	High risk	Randomisation led to differences in sex distribution with 79% men in the treatment group versus 32% men in the placebo group.
Compliance	Unclear risk	This was not described.
Severity of AD	Low risk	This was rated as moderate and severe.
Conflict of interest	Unclear risk	No details were given.