Repace and Lowrey
J.L. Repace and A.H. Lowrey attempted to derive the number of deaths attributable to passive smoking as measured by various different techniques. The contents of their paper A quantitative estimate of nonsmokers’ lung cancer risk from passive smoking were ridiculed by Burch in a critique that can hardly be improved on. What follows is quotation from Burch.
Outline of Repace and Lowrey’s Procedure
1. Exposure of non-smokers
From measurements of the concentration of respirable particles between 0.01 and 3.0 μm diameter in representative samples of room air, and from surveys of lifestyles. Repace and Lowrey (1985) estimate the total exposures of nonsmokers (range and average) in terms of mg cigarette tar per day.
2. Epidemiologic studies of lung cancer in passive smokers
They review 13 epidemiologic studies of lung cancer risk in the nonsmoking spouses of cigarette smokers. In 12 of these, the only index of exposure was the strength of the spouse’s smoking habit. The mortality ratio for persons married to smoking spouses versus those married to nonsmokers clusters around a value of 2 .0. They also compare the sex- and age-adjusted mortality from lung cancer in Seventh Day Adventists (SDA) self reported nonsmokers with that in never-smokers in the general U.S. population. Non-SDA never-smokers had an average standardised death rate from lung cancer that was 2.4 times that of SDA never-smokers. This last group, they believe, was less likely to be exposed to ambient tohacco smoke than the former.
3. Estimated death rate from lung cancer in nonsmokers using epidemiologic studies of passive smoking
Repace and Lowrey (1985) assume that the mortality ratio (2.4) derived from the SDA study implies causation and thereby calculate that some 4700 lung cancer deaths per year have been caused among U.S. nonsmokers, owing to passive smoking. They conclude that this figure agrees well (within 10%) with Hirayama’s (1981) estimate of mortality from lung cancer attributable to passive smoking by the non smoking wives of Japanese smokers.
4. Estimated death-rate from lung cancer in nonsmokers using epidemiologic studies of active smokers
From many studies of lung cancer in active cigarette smokers reviewed by the U.S. Surgeon General (1982), and a linear exposure-response relationship, Repace and Lowrey derive an alternative estimate of the lung cancer death rate in the U .S. resulting from passive smoking. This procedure leads to an estimated 555 lung cancer deaths per year, nearly one order of magnitude lower than the rate (4700 per year) calculated under procedure 3 above.
5. Discussion of discrepancy between procedures 3 and 4.
Repace and Lowrey speculate upon ways in which the lower estimate might be raised to the upper value.
Burch proceeded to comment on each step in turn, emphasising the difficulty of estimating exposure to tobacco smoke and the inadequacy of spouse’s smoking as a proxy. Passive smoking studies up to this point had mostly studied exposure in the home to the exclusion of the workplace. About this time, urine or saliva tests for levels of the nicotine metabolites cotinine and creatinine were introduced, making the use of spousal smoking redundant. As for step 5:
On encountering a near order of magnitude discrepancy between alternative estimates, most investigators would express dismay: it is to the credit of Repace and Lowrey that they display no outward signs of disappointment. Nevertheless, their “…discussion of alternative exposure-response relationships” has one conspicuous feature. Means are sought only to raise the lower estimate of 555 lung cancer deaths per years to the higher (4700 lung cancer deaths per year) and none in the reverse direction.