Nasal tract, pharynx and pulmonary tumours Meta-analyses

11.6.1 Nasal tract, pharynx and pulmonary tumours Meta-analyses

Blair et al. (1990a) conducted a meta-analysis of 32¹ studies covering occupational exposure to formaldehyde in industrial workers and professionals (embalmers, anatomy technicians and pathologists). The data were re-analysed by Partanen (1993) and included an additional three case-control studies¹. Furthermore, in the meta-analysis by Partanen (1993) a number of changes in the selection of input values were made that were considered more appropriate, and relative risks determined using a different model from that of Blair et al. (1990a). Despite these changes the results of this re-analysis were generally in close agreement with the original meta-estimates by Blair et al. (1990a).

respectively). The two meta-analyses showed no increased risk between formaldehyde exposure and lung cancer among professionals. The mRR for lung

1. 
   Brinton et al., 1984; Gallagher et al., 1986; Merletti et al., 1991.
   

cancer for industrial workers was marginally, but significantly, increased for those with low/low-medium exposure to formaldehyde (both mRR = 1.2, 95% CI

1. 
   - 1.3), but a significantly increased risk was not observed in both meta- analyses for those exposed to higher/substantial levels of formaldehyde. The observed marginally increased risk in the low dose group in the absence of a dose response does not demonstrate strong evidence of an association between formaldehyde exposure and lung cancer. For nasal cancers, Blair et al. (1990a) found no increased risk for formaldehyde exposure overall, while Partanen (1993) found a borderline significantly increased risk of sinonasal cancers in workers with substantial exposure to formaldehyde (mRR = 1.7, 95% CI 1.0 - 2.8).
   

In a more recent and comprehensive meta-analysis, Collins et al. (1997) initially considered 47 epidemiology studies. Several of these studies were not included in the analysis, because workers who had formaldehyde exposure were not evaluated separately or the study only reported relative risks, the study population was included in a more recent study, or the methodology and results were insufficiently described. In total¹ the meta-analysis was based on the results from 11 cohort, 3 proportionate mortality and 18 case-control studies, and included new data published since Partanen (1993). Furthermore, the authors of studies were contacted to obtain data not included in their publications. The exposure potential of jobs that were classified as having formaldehyde exposure in the community-based case-control studies was also reviewed, as exposure assessment was much more uncertain in these studies than in cohort studies.

0.9 - 1.1). Similarly, no increased risk was seen in the case-control studies (mRR

= 0.8, 95% CI 0.7 - 0.9).
No increased risk of sinonasal cancers was seen with exposure to formaldehyde (mRR = 1.0, 95% CI 1.0 - 1.1). Evaluating by study design revealed no increased risk for cohort studies (mRR = 0.3, 95% CI 0.1 - 0.9) but a significantly increased risk for case-control studies (mRR = 1.8, 95% CI 1.4 - 2.3). This increased risk was attributable to a significantly increased risk for the combined 6 European case-control studies (mRR = 2.9, 95% CI 2.2 – 4.0), whereas no increased risk was seen for the combined 5 US case-control studies (mRR = 1.0 95% CI 0.7 - 1.5). Collins et al. (1997) report that it is difficult to reconcile European findings with other findings unless it is assumed that confounding factors, or bias, were affecting the results.

¹ Harrington and Shannon, 1975*; Jensen and Andersen, 1982*; Fayerweather et al., 1983*: Hernberg et al., 1983a; 1983b; Walrath and Fraumeni, 1983*; Coggon et al., 1984*; Levine et al., 1984*; Walrath and Fraumeni, 1984*; Brinton et al., 1985; Bond et al., 1986*; Bertazzi et al., 1989*; Blair et al., 1986*; Hayes et al., 1986a*; Olsen et al., 1986; Stroup et al., 1986*; Vaughan et al., 1986a*; 1986b*; Roush et al., 1987*; Stayner et al., 1988*; Gerin et al., 1989*; Hayes et al., 1990*; Partanen et al., 1990; Hall et al., 1991; Matanoski, 1991; Chiazze et al., 1993; Gardner et al.,1993; Luce et al., 1993; West et al., 1993; Marsh et al., 1994; Andjelkovich et al., 1995 (* included in the analysis by Blair et al., 1990a and Partanen, 1993).

men, 102 women) and 195 adenocarcinomas (169 men, 26 women), diagnosed

between 1968 and 1990 were evaluated along with 3136 controls (2349 men, 787 women). The probability of exposure to a number of occupational substances (including formaldehyde) was determined using a job exposure matrix. The study focused on cumulative exposure although results of other exposure variables were presented when they gave additional information. After adjustment for age, a small non-significant increased risk was seen for squamous cell carcinomas in males and females with a high probability of exposure (odds ratio (OR) = 1.2, 95% CI 0.8 – 1.8 and OR = 1.5, 95% CI 0.6 – 3.8, respectively for a > 90% probability of exposure). After adjustment for age and cumulative exposure to wood and leather dust a significantly increased risk was seen between adenocarcinomas and medium (0.25 - 1 ppm) and high (> 1 ppm) intensity of exposure to formaldehyde in men (OR = 2.4, 95% CI 1.3 - 4.5 and OR = 3.0, 95% CI 1.5 - 5.7, respectively). Only age was adjusted for in women, with a significantly increased risk seen between adenocarcinomas and high probability of formaldehyde exposure (OR = 6.2, 95% CI 2.0 - 19.7).
¹ Cecchi et al., 1980, Luce et al., 1993 and Leclerc et al., 1994; Hardell et al., 1982; Brinton et al., 1984 and Brinton et al., 1985; Merler et al., 1986; Hayes et al., 1986a and Hayes et al., 1986b; Vaughan et al., 1986a, Vaughan, 1989 and Vaughan and Davis, 1991; Bolm-Audorff et al., 1990; Comba et al., 1992a; Comba et al., 1992b; Zheng et al., 1992; Magnani et al., 1993; Mack and Preston-Martin unpublished data, presented in Luce et al., 2002.
Luce et al. (2002) also evaluated cases of sinonasal adenocarcinoma where there was no exposure to wood or leather dust. A significantly increased risk was only seen for adenocarcinoma in females with a high probability of exposure (OR = 11.1, 95% CI 3.2 – 38.0, based on 5 cases). No significant increased risk was seen in males for low, medium or high probability of exposure. An analysis was also undertaken in men only of formaldehyde exposure by maximum exposure to wood dust. For no or low exposure to wood dust a non-significant increased risk was seen for adenocarcinomas with high and medium level exposure to formaldehyde (mRR = 2.2, 95% CI 0.8 – 6.3 based on 4 cases).