Appendix C:
Adjustments to Epidemiologic Estimates of Excess Lung Cancer in Persons Exposed to Environmental Tobacco Smoke
Chapter 12 describes 13 epidemiologic studies that estimate the relative risks of lung cancer in nonsmoking spouses of smokers compared with nonsmoking spouses of nonsmokers. A weighted average of the relative risks of “exposed” to “unexposed” persons is 1.34, i.e., a 34% increase in lung cancer risk as a consequence of environmental tobacco smoke (ETS) exposure. On the other hand, one can extrapolate in a linear fashion from the relative levels of cotinine that had been measured in active smokers and exposed nonsmokers. The expected relative risk for exposed nonsmokers would range from about 1.03 to 1.10. Neither of these estimates has been corrected for possible misclassification of subjects in the epidemiologic studies. The latter risk assumes that the onetime measure is a satisfactory surrogate for lifetime exposure. Misclassification problems and problems in estimating actual carcinogen exposure make it very difficult to provide an estimate of the numbers of cancer cases both in smokers and nonsmokers that might be attributable to ETS.
In this section we combine information from several sources to generate crude estimates of the relative risk to nonsmokers as a consequence of chronic exposure to ETS. The computations reported here are highly simplified and should be looked on as providing only a first approach to risk evaluation. A more detailed approach, including a more explicit statement of the assumptions involved, is given in Appendix D. A major concern is that persons who have been identified as “unexposed” to ETS may have really been exposed. If this were true, then the risks relative to truly unexposed persons would be underestimated. To estimate this
possible effect, the results from studies of urinary cotinine are used here to adjust for the proportion of selfreported “unexposed” nonsmokers who, in fact, may have been exposed to ETS.
USING COTININE MEASUREMENTS TO CORRECT MISREPORTING
The only source of cotinine or nicotine in body fluids is tobacco smoke exposure. Therefore, urinary cotinine provides an objective measure of (recent) exposure. It has been reported (Jarvis et al., 1984; Wald et al., 1984) that urinary nicotine and cotinine are 3 times as high in “exposed” nonsmoking spouses of current smokers than in “unexposed” nonsmoking spouses of current nonsmokers. For example, Wald and Ritchie (1984) report urinary cotinine in the ratio 1:3:215 for “unexposed” nonsmokers, ETSexposed nonsmokers, and regular smokers, respectively.
Several assumptions need to be made to permit the use of these data before any quantitative risk computation can be made:

Current smoking patterns reflect past patterns.

Cotinine or nicotine concentrations in the urine are linearly related to recent exposures to ETS and to the carcinogens in ETS among nonsmokers.

All subjects in the various studies began to be exposed to ETS at the same age and have continued to be exposed at the same rate throughout the followup period.

The excess relative risk for lung cancer in nonsmokers is proportional to the dose (in cigarette equivalents) of ETS absorbed.
An assumption of a linear doseresponse relationship implies that if the risk (i.e., mortality rate) at a given age (t) for a specific calendar period(s), given some absorbed dose (d), then γ(t,sd)
(1)
This equation expresses the risk as equal to the base mortality risk, γ_{0}(t,8), for a truly unexposed person for the same age and calendar period, multiplied by an excess relative risk that increases linearly with dose, i.e. (1+βd), where β is the amount of increase
per unit dose.* Further, the risk for a truly unexposed nonsmoker, i.e., γ_{0}(t,s), is assumed to be the same for men and women. This assumption is supported, in part, by the results given in Chapter 12 and, in part, by earlier studies of Garfinkel (1981) and Friedman et al. (1984). Doll (1984), however, gives different risks for men and women of lung cancer mortality in nonsmokers.
If d_{E} is actual dose in the “exposed” persons and d_{N} is the actual dose in persons who believe themselves to be “unexposed,” then we have, from Equation 1:
(2A)
and
(2B)
The relative risk for a person identified as “exposed” compared to a person identified as “unexposed” [RR(d_{E})] is given by Equation 2A divided by 2B:
(3)
which, from Chapter 12, is 1.34, the relative risk estimated from the epidemiologic studies.
From the studies that measured cotinine in “exposed” and “unexposed” persons, we assume that the operative dose level, d_{E}, among “exposed” individuals is 3 times as high as the dose level in the selfreported “unexposed” persons, d_{N}, and that the ratio of 3:1 is proportional to a lifetime dose difference. Therefore, Equation 3 may be rewritten as:
(4)
Equation 4 can be solved for βd_{N}, which is the increase in risk for persons called “unexposed,” but who, in fact, have been exposed
to some recent ETS, as indicated by their nonzero levels of urinary cotinine or nicotine. Solving Equation 4 gives:
Thus, the relative risk for a selfidentified “unexposed” person compared with a truly unexposed person is:
and the relative risk for an “exposed” person compared with a truly unexposed person is:
To see what possible effect these relative risk estimates would have on the populationattributable risk, i.e., the fraction of lung cancer in nonsmoking individuals attributable to ETS, the proportion of the population that is exposed to ETS needs to be estimated. Wald and colleagues (1984) have reported that 17% of nonsmoking women and 12% of nonsmoking men fall into the category of “exposed,” i.e., nonsmoking spouses of smokers. By subtraction, this means that 83% of nonsmoking women and 88% of nonsmoking men would consider themselves “unexposed.” Given this, we can estimate the populationattributable risk, which is given in general form as:
(5)
where p_{1} is the proportion of people who call themselves “exposed,” RR_{1} is the relative risk of selfreported “exposed” persons, and RR_{2} is the relative risk of selfreported “unexposed” persons. Thus, for men:
and for women:
That is, about 21% of the lung cancers in nonsmoking women and 20% in nonsmoking men may be attributable to exposure to ETS.
REFERENCES
Doll, R.D. Epidemiological discovery of occupational cancers. Ann. Acad. Med. Singapore 13(Suppl.):331–339, 1984.
Doll, R.D., and R.Peto. Cigarette smoking and bronchial carcinogenic: Dose and time relationships among regular smokers and lifelong nonsmokers. J. Epidemiol. Comm. Health 32:303–313, 1978.
Friedman, G.D., R.D.Bawol, and D.B.Pettiti. Prevalence and correlates of passive smoking. Am. J. Public Health 73:401–405, 1983.
Garfinkel, L. Time trends in lung cancer mortality among nonsmokers and a note on passive smoking. J. Natl. Cancer Inst. 66:1061–1066, 1981.
Jarvis, M., H.TunstallPedoe, C.Feyerabend, C, Vesey, and Y.Salloyee. Biochemical markers of smoke absorption and self reported exposure to passive smoking. J. Epidemiol. Comm. Health 38:335–339, 1984.
Wald, N.J., A.Boreham, A.Bailey, C.Ritche, J.E.Haddow, and G.Knight. Urinary cotinine as a marker of breathing other people’s smoke. Lancet 1:230–231, 1984.
Wald, N.J., and C.Ritchie. Validation of studies on lung cancer in nonsmokers married to smokers. Lancet 1:1067, 1984.