Health Assessment Document for Diesel Emissions - NSCEP | US ...

1 from surrogates (next of kin being either a spouse or an offspring) were found to be accurate by
2 Lerchen and Samet (1986) and McLaughlin et al. (1987). Lerchen and Samet (1986) did not
3 detect any consistent bias in the report of cigarette consumption. In contrast, overreporting of
4 cigarette smoking by surrogates was observed by Rogot and Reid ( 197 5), K.olonel et al. ( 1977),
5 and Humble et al. (1984). Kolonel et al. (1977) found that the age at which an individual started-
6 smoking was reported within 4 years of actual age 84% ofthe time. The indication from these
7 studies is that surrogates were able to provide fairly credible information on the smoking habits
8 of the study subjects. If the surrogates of the cases were more likely to. overreport cigarette
9 smoking as compared to the controls, then it might be harder to find an effect of diesel exhaust
1 0 because most of the increase in lung cancer would be attributed to smoking rather than to the
11 effect of exposure to diesel exhaust.
12
13 8.5.5. Criteria of Causal Inference
14 In most situations epidemiologic data are used to delineate the causality of certain health
15 effects. Several cancers have been causally associated with exposure to agents for which there is
16 no direct biological evidence. Insufficient knowledge about the biological basis for diseases in
17 humans makes it difficult to identify exposure to an agent as causal, particularly for malignant
18 diseases when the exposure was in the distant past. Consequently, epidemiologists and biologists
19 have provided a set of criteria that define a causal relationship between exposure and the health
·20 outcome. A causal interpretation is enhanced for studies that meet these criteria. None of these
21 criteria actually proves causality; actual proof is rarely attainable when dealing with
22 environmental carcinogens. None of these criteria should be considered either necessary (except
23 temporality of exposure) or sufficient in itself. The absence of any one or even several of these
24 criteria does not prevent a causal interpretation. However, if more criteria apply, it provides
25 more credible evidence for causality.
26 Thus, applying the criteria of causal inference to the seven cohort mortality and eight
. .
27 case-control stuqies in which risk of lung cancer was assessed resulted in the following:
28
29
30"
31
32
33
34
35
36
37
2/1198
• Temporality: There is a temporality of exposure to diesel exhaust prior to the
occurrence of lung cancer.
• Strength of association: The strength of association between exposure and the
occurrence of lung cancer in the cohort studies showed a 30% to 57% higher risk
among exposed persons as compared to nonexposed (Howe et al., 1983; Wong et al.,
1985; Boffetta and Stellman, 1988; Garshick et al., 1988). In case-control studies, the
risk varied from 20% to 89% higher among exposed as compared to nonexposed
(Williams et al., 1977; Hall and Wynder, 1984; Damber and Larsson, 1987; Garshick
8-66 DRAFT--DO NOT CITE OR QUOTE