Exposure-response relationships for select cancer and non-cancer health outcomes in a cohort of U.S. firefighters from San Francisco, Chicago and Philadelphia (1950-2009)

Abstract

Objectives: To examine exposure-response relationships between surrogates of firefighting exposure and select outcomes among previously studied US career firefighters.

Methods: Eight cancer and four non-cancer outcomes were examined using conditional logistic regression. Incidence density sampling was used to match each case to 200 controls on attained age. Days accrued in firefighting assignments (exposed-days), run totals (fire-runs) and run times (fire-hours) were used as exposure surrogates. HRs comparing 75th and 25th centiles of lagged cumulative exposures were calculated using loglinear, linear, log-quadratic, power and restricted cubic spline general relative risk models. Piecewise constant models were used to examine risk differences by time since exposure, age at exposure and calendar period.

Results: Among 19,309 male firefighters eligible for the study, there were 1333 cancer deaths and 2609 cancer incidence cases. Significant positive associations between fire-hours and lung cancer mortality and incidence were evident. A similar relation between leukaemia mortality and fire-runs was also found. The lung cancer associations were nearly linear in cumulative exposure, while the association with leukaemia mortality was attenuated at higher exposure levels and greater for recent exposures. Significant negative associations were evident for the exposure surrogates and colorectal and prostate cancers, suggesting a healthy worker survivor effect possibly enhanced by medical screening.

Conclusions: Lung cancer and leukaemia mortality risks were modestly increasing with firefighter exposures. These findings add to evidence of a causal association between firefighting and cancer. Nevertheless, small effects merit cautious interpretation. We plan to continue to follow the occurrence of disease and injury in this cohort.

Figure 1

Relative risk of lung cancer (A) and leukaemia (B) mortality by career fire-runs. Models indicated by: solid line, loglinear; long dashed line, linear; short dashed line, log-quadratic; long dash followed by two dots, power; solid line with filled circles, RCS with knots indicated by filled circles. Vertical lines indicate 25th, 50th and 75th centiles of exposure. Plot truncated at the 95th centile. The shaded area shows the 95% CI for the preferred model (loglinear for lung cancer and power for leukaemia). RCS, restricted cubic splines.