© 2004 by Oxford University Press
© 2004 Oxford University Press
CORRESPONDENCE |
RESPONSE: Re: Extended Follow-up of a Cohort of British Chemical Workers Exposed to Formaldehyde
Affiliation of authors: MRC Environmental Epidemiology Unit, University of Southampton, Southampton, U.K.
Correspondence to: Professor David Coggon, MRC Environmental Epidemiology Unit, Southampton General Hospital, Southampton SO16 6YD, U.K. (e-mail: dnc{at}mrc.soton.ac.uk)
As Dr. Greenberg points out, the optimal choice of reference rates in occupational cohort studies is not always straightforward. The advantage of comparison with data for the general population is that these data are often readily available and statistically stable. Bias may occur, however, if a substantial proportion of the reference population is materially exposed to the agent or agents under study or if they differ importantly in their exposure to other risk factors for the disease.
In our study, we calculated expected numbers of deaths for lung cancer from national rates, both with and without adjustment for local differences in mortality. Neither method is ideal. On the one hand, comparisons with the national population are potentially more prone to confounding by differences in smoking habits and in exposure to outdoor air pollutants. On the other hand, the local populations whose mortality was used to adjust expected numbers will have included proportionately more cohort members and therefore a higher percentage of individuals with exposure to formaldehyde. This proportion was still relatively small, however, and we therefore give greater weight to the locally adjusted analysis, although as we indicated, there remains a possibility of residual confounding.
Greenberg also raises the possibility of carcinogenic exposures in other industries near our study factories, a point that has been made by others (1). The potential for confounding would then depend on the extent to which cohort members had at some time worked in the industries concerned. If this were common, the case for a locally adjusted comparison would be even stronger. If it were rare, a national comparison might have advantages.
Adjusting for social class would be another way of addressing nonoccupational confounders. However, we preferred geographical adjustment because place of residence seemed likely to provide a better proxy for relevant confounders (smoking and outdoor air pollution) and had been used in both previous analyses of the cohort (2,3).
The acceptability of risk entails value judgments that will differ from person to person and according to the nature of the hazard and the perceived benefits that accompany the risk. We agree that no epidemiologic study can directly exclude an excess lifetime risk of one in 100 000, except for an extremely rare disease. Nevertheless, epidemiology does contribute usefully to risk assessment, even when such low risks are of concern. An absence of detectably elevated risk in an adequately powered study of heavily exposed workers can provide strong reassurance that any risks from much lower exposures in the general environment will be negligible. Also, when exposure limits are set on the basis of toxicologic data, there will inevitably be uncertainties associated with the extrapolation from animals to man. It is therefore important to check, if possible, that epidemiologic data do not indicate a higher risk than would be predicted from animal model data.
In the case of formaldehyde, the available epidemiology is broadly reassuring, but additional data from cohorts such as ours could give greater reassurance or, alternatively, might lead to the identification of a hazard that is at present unclear.
REFERENCES
1 Infante PF, Schneiderman MA. Formaldehyde, lung cancer and bronchitis. Lancet 1986;1:436–7.[Medline]
2 Acheson ED, Gardner MJ, Pannett B, Barnes HR, Osmond C, Taylor CP. Formaldehyde in the British chemical industry. Lancet 1984;1:611–6.[Web of Science][Medline]
3 Gardner MJ, Pannett B, Winter PD, Cruddas AM. A cohort study of workers exposed to formaldehyde in the British chemical industry: an update. Br J Ind Med 1993;50:827–34.[Web of Science][Medline]
Related Correspondence
CiteULike 
Connotea 
Del.icio.us What's this?
J Natl Cancer Inst 2004 96: 1037.
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||