Table 3

Within-subject effect of day (welding/classroom area) and time (morning/afternoon) on three metabolites, as well as overnight effects

Sphingosine 1-phosphateSphinganine 1-phosphateSphingosine
Mean* (SD)Mean (SD)Mean (SD)
Two-factor factorial ANOVAPr>FPr>FPr>F
N/AM0.24 (0.33)0.16 (0.44)0.25 (0.63)
N/PM−0.04 (0.33)−0.15 (0.49)−0.13 (0.49)
W/AM−0.16 (0.34)Pinteraction<0.01−0.27 (0.43)Pinteraction<0.01−0.18 (0.69)Pinteraction=0.06
W/PM−0.16 (0.26)−0.05 (0.52)−0.29 (0.65)
Previous afternoon and welding dayPr>|t|
Previous PM0.02 (0.36)†<0.31−0.04 (0.49)0.720.07 (0.78)0.01
W/AM−0.13 (0.28)<0.01−0.03 (0.41)0.24−0.18 (0.86)0.91
W/PM−0.02 (0.29)−0.09 (0.41)−0.17 (0.58)
Welding day and next morning
W/AM0.06 (0.44)<0.010.00 (0.61)<0.01−0.04 (0.46)<0.01
W/PM−0.27 (0.28)0.39−0.24 (0.42)0.01−0.29 (0.38)0.42
Next morning−0.17 (0.29)−0.19 (0.48)−0.17 (0.40)
  • The sample sizes for two-factor factorial ANOVA, pre-exposure overnight design and postexposure overnight design were 176, 297 and 132, respectively.

  • Pr>F indicates probability that the predictor has no effect on the outcome variable.

  • Pr>|t| indicates probability that the level was different compared with the last measurement.

  • Interaction indicates the effect between the previous two factors.

  • *The mean of metabolites was calculated from concentrations after normalisation, log transformation and mean centring.

  • †These values are for both current smokers and non-current smokers.

  • AM/PM, afternoon compared with morning; ANOVA, analysis of variance; interaction, interaction effect between the previous two factors; W/N, welding day compared with non-welding day.