# Table 2 Relative risk of hospital-treated pneumonia by vitamin E supplementation in participants initiating smoking at ≤ 20 years, ATBC Study 1985–1993

Intervention Effect of vitamin E
Vitamin E No vitamin E
Subgroup No. of participants No. of cases Rate* No. of cases Rate* RR (95% CI)*
All 21657 370 6.0 331 5.3 1.14 (0.98–1.32)
Weight (kg) †
36–59 1054 47 16.7 32 10.4 1.61 (1.03–2.53)
60–69 4115 79 6.7 68 5.7 1.17 (0.84–1.62)
70–89 12495 182 5.1 187 5.1 0.99 (0.81–1.22)
90–99 2653 39 5.2 34 4.5 1.17 (0.74–1.86)
100–154 1328 22 5.7 9 2.5 2.34 (1.07–5.08)
β-Carotene
No 10842 204 6.6 169 5.4 1.22 (0.99–1.50)
Yes 10815 166 5.4 162 5.2 1.04 (0.84–1.30)
1. * Rate is given as the number of pneumonia cases per 1,000 person-years. Risk ratio (RR) was estimated using the proportional hazards regression model comparing participants who received vitamin E and those who did not. No covariates were included in the models, because the comparison is between large randomized groups. The sizes of the compared intervention groups are the same within 10% accuracy in the lowest and highest body-weight groups, and within 2% in the other comparisons. The uniformity of the vitamin E effect was tested by adding a dummy variable for vitamin E effect in the 36–59, 60–69, 90–99 and 100–154 kg groups, allowing each of the four groups their own vitamin E supplementation effect. The regression model was improved by χ2(4 df) = 7.76, P = 0.10, compared to the model with a uniform vitamin E effect. Adding the vitamin E effect only in the 36–59 and 100–154 kg groups improved the model by χ2(2 df) = 6.79, P = 0.034, compared to the model with a uniform vitamin E effect.
2. † Weight was missing for 12 participants; two of them had pneumonia, one in the vitamin E group and one in the no-vitamin E group.