Table 3.
Full sample | Stratified by smoking status* | |||||
---|---|---|---|---|---|---|
All treatment groups: | Unadjusted | Adjusted (model-based) | P value a | Never smokers | Former smokers | Current smokers |
Mean (SD) † | Mean (95% CI) †† | Mean (95% CI) †† | Mean (95% CI) †† | Mean (95% CI) †† | ||
Placebo | −41 (74) | −39 (−46, −32) | reference | −40 (−52, −29) | −35 (−45, −25) | −51 (−67, −35) |
Vitamin E | −34 (75) | −33 (−40, −26) | 0.1866 | −31 (−42, −20) | −36 (−46, −26) | −29 (−44, −14) |
Selenium | −37 (74) | −38 (−44, −31) | 0.7502 | −37 (−48, −26) | −38 (−48, −28) | −38 (−54, −23) |
Vitamin E and Selenium | −34 (74) | −34 (−41, −28) | 0.3133 | −32 (−44, −20) | −32 (−41, −22) | −44 (−59, −29) |
Marginal Models b : | ||||||
Any Selenium versus Placebo | Placebo | −39 (−46, −32) | reference | −40 (−52, −29) | −35 (−45, −25) | −49 (−66, −33) |
Any Selenium | −36 (−41, −31) | 0.4442 | −35 (−43, −27) | −35 (−42, −28) | −42 (−52, −31) | |
Any Vitamin E versus Placebo | Placebo | −39 (−42, −36) | reference | −40 (−51, −29) | −34 (−44, −24) | −49 (−66, −33) |
Any Vitamin E | −34 (−38, −30) | 0.1772 | −31 (−40, −23) | −34 (−41, −27) | −37 (−47, −26) |
Abbreviations: FEV 1 forced expiratory volume in the first second, mL milliliter.
*All smoking-stratified treatment vs. placebo comparisons exceeded the P value threshold of 0.05 (all values ≥ 0.25); in current smokers vitamin E vs. placebo P = 0.3552, selenium vs. placebo P = 0.4985, vitamin E + selenium vs. placebo P = 0.8887, any selenium vs. placebo P = 0.8705.
†Mean decline (SD); calculated for men with ≥2 years between first and last pulmonary function tests where date of last test ≤ March 2009; calculated in 418 (placebo), 426 (vitamin E), 416 (selenium) and 417 (vitamin E + selenium) participants.
††Model-based estimated mean decline and 95% confidence interval (Additional file 1: Tables S3 and S4 for regression model results); main effects models adjusted age, height, race, smoking status, time and tested time x treatment effect (effect of treatment on slope); smoking interaction models adjusted age, height, race, smoking status, and all two way interactions to test smoking x time x treatment interaction (effect of treatment on slope differs by smoking).
aP value from mixed-effects linear regression model, with placebo reference group.
bMarginal models combine treatment groups. Any selenium model compares participants on any selenium (selenium alone and in combination with E) to placebo. In a separate model, participants on any vitamin E (vitamin E alone and in combination with selenium) are compared to placebo.