Table 3.

 Distributions of primary and two secondary outcome measures in each group and difference between groups. Values are means (95% confidence intervals) unless stated otherwise

Outcome measures	Intervention group*	Control group	Difference between means (95% CI)
			Unadjusted parametric analysis	P for difference	Adjusted† parametric analysis	P for difference
HECSI score	4.97 (4.14 to 5.88)	8.53 (7.45 to 9.63)	−3.56 (−4.92 to −2.14)	<0.001	−3.47 (−4.80 to −2.14)	<0.001
DLQI score	1.22 (0.88 to 1.61)	2.00 (1.58 to 2.48)	−0.78 (NA)	0.003‡	−0.92 (−1.48 to −0.37)	<0.001
Score of knowledge of hand eczema and skin protection	10 (1)	10 (0)	—	0.42§	—	NA

HECSI=hand eczema severity index; DLQI=dermatology life quality index; NA=not applicable.

Non-parametric test (Mann Whitney) was used if assumptions of the general linear univariate model were not fulfilled.

*Values were square root transformed before calculations. Means (95% CI) were then squared.

†Adjusted by stratification variables and baseline values.

‡Standardised residuals were not normally distributed (Shapiro Wilk’s test P<0.001 and distribution significantly right skewed). Therefore the non-parametric Mann Whitney test was used and the P value of that test is shown. By contrast when the analysis was adjusted the fit of the general linear model was perfect.

§Non-parametric test used (Mann Whitney) at it was not possible to transform quantities for acceptable model fit.