Table 2.

Correction of NCI 1994 data and additional follow-up for leukemia and resulting attenuated exposure-response relationships.

			NCI study
	1994 follow-up originala,b		1994 follow-up revisedc		2004 follow-upd		Marsh et al. (2004) re-analysise
Highest peak(ppm)	No. deaths	RR (95% CI)	No. deaths	RR (95% CI)	No. deaths	RR (95% CI)	SMR (95% CI)
Unexposed	4	0.78(0.25-2.43)	4	0.52(0.17-1.57)	7	0.59(0.25-1.36)	0.38* (0.10-0.97)
>0-1.9(base)	16	1.00 (—)	23	1.00 (—)	41	1.00 (—)	0.50* (0.28-0.81)
2.0-3.9	20	2.04(1.0-4.01)	20	1.36(0.73-2.51)	27	0.98(0.60-1.62)	1.04(0.63-1.60)
>4.0	29	2.46(1.31-4.62)	29	1.60(0.90-2.82)	48	1.42(0.92-2.18)	1.31(0.88-1.89)

^aUsed by IARC in 2004;

^bp_trend = .001 (all groups), p_trend = .004 (exp. only);

^cp_trend = .021 (all groups), p_trend = .094 (exp. only)

^dp_trend = .02 (all groups), p_trend = .12 (exp. only)

^ep < 0.05.