Table 3

Model goodness of fit and association of lung cancer risk with alternative indicators of cumulative exposure to asbestos fibres, estimated by Poisson regression with adjustment for age, sex, race, calendar time and birth cohort

Exposure indicator	β	SE (β)	∆	LR (df)	p	AIC
Biologically-based indices^*
Modified Lippmann Index (0.25≤D≤1 μm and L≥10 μm)^†	0.00283	0.00073	0.0279	11.1 (1)	0.001	2341.6
Berman Index	0.01880	0.00551	0.0326	8.8 (1)	0.003	2343.9
TEM fibres (D<0.25 μm and L≥10 μm)	0.00171	0.00054	0.0321	8.3 (1)	0.004	2344.4
Ad hoc indices (μm-y)
Mean fibre length	0.00869	0.00257	0.0757	9.7 (1)	0.002	2343.1
Mean fibre diameter	0.15098	0.04558	0.0726	9.2 (1)	0.002	2343.5
Mean fibre length	0.01919	0.02400	0.1329	9.8 (2)	0.007	2344.9
Mean fibre diameter	−0.18835	0.42987	−0.0735

↵* Units are f-y/ml. All of the indices also include the criterion that the aspect ratio (length:diameter) is at least 3:1.
↵† Lippmann proposed minimum diameters of 0.15 μm or 0.3 μm, but we used a cut-off of ≥0.25 μm since that was the closest category in our TEM protocol.
AIC, Akaike Information Criterion (smaller values indicate better fit); β and SE, regression coefficient and associated standard error; ∆, change in lung cancer risk for increment in exposure equal to 1 IQR; D, diameter; L, length; LR (df), likelihood ratio test statistic (equivalent to χ²) with degree of freedom (df) and associated p value; TEM, transmission electron microscopy.