Long-term effects of traffic-related air pollution on mortality in a Dutch cohort (NLCS-AIR study)

Abstract

Background: Several studies have found an effect on mortality of between-city contrasts in long-term exposure to air pollution. The effect of within-city contrasts is still poorly understood.

Objectives: We studied the association between long-term exposure to traffic-related air pollution and mortality in a Dutch cohort.

Methods: We used data from an ongoing cohort study on diet and cancer with 120,852 subjects who were followed from 1987 to 1996. Exposure to black smoke (BS), nitrogen dioxide, sulfur dioxide, and particulate matter < or = 2.5 microm (PM(2.5)), as well as various exposure variables related to traffic, were estimated at the home address. We conducted Cox analyses in the full cohort adjusting for age, sex, smoking, and area-level socioeconomic status.

Results: Traffic intensity on the nearest road was independently associated with mortality. Relative risks (95% confidence intervals) for a 10-microg/m(3) increase in BS concentrations (difference between 5th and 95th percentile) were 1.05 (1.00-1.11) for natural cause, 1.04 (0.95-1.13) for cardiovascular, 1.22 (0.99-1.50) for respiratory, 1.03 (0.88-1.20) for lung cancer, and 1.04 (0.97-1.12) for mortality other than cardiovascular, respiratory, or lung cancer. Results were similar for NO(2) and PM(2.5), but no associations were found for SO(2).

Conclusions: Traffic-related air pollution and several traffic exposure variables were associated with mortality in the full cohort. Relative risks were generally small. Associations between natural-cause and respiratory mortality were statistically significant for NO(2) and BS. These results add to the evidence that long-term exposure to ambient air pollution is associated with increased mortality.

Keywords: air pollution; cohort; mortality; traffic.

Figure 1

Distribution of estimated NO₂ (background and overall estimate), BS (background and overall estimate), SO₂ (background), and PM_2.5 (overall estimate) concentrations (1987–1996), and of the traffic intensity on the nearest road and the sum of traffic intensity in a 100-m buffer, at the 1986 home address (n = 117,528). Abbreviations: Max, maximum; mi, minimum.

Figure 2

Adjusted results of spatial analyses for association between cardiopulmonary mortality and BS background concentration (1987–1996) (A) and traffic intensity on the nearest road in the full cohort (n = 107,005) (B). RRs and 95% CIs are shown for the original, 1-level neighborhood independent-clusters (analysis a), 1-level municipality independent-clusters (analysis b), 2-level independent-clusters (analysis c), 1-level neighborhood distance-decay (analysis d), 1-level municipality distance-decay (analysis e), and 2-level distance-decay (analysis f) analyses (confounders used are age, sex, smoking status, and area-level indicators of socioeconomic status).

Figure 3

Association between black smoke overall concentration (1987–1996) and cause-specific mortality in subgroups for cigarette smoking status in the full cohort data set (A–E), and (F) by education and fruit consumption in the case–cohort data set. (A) Natural-cause (p = 0.15), (B) cardiovascular (p > 0.2), (C) respiratory (p = 0.11), (D) lung cancer (p = 0.14), and (E) other mortality (p > 0.2). (F) Education of the household coded as low = only primary school; middle = lower vocational education; and high = junior high school, senior high school, higher vocational education, and university (p > 0.2). Fruit consumption divided in tertiles: low, 0–96.8 g/day; medium, 96.8–191.8 g/day; and high, > 191.8 g/day. Adjusted for age, sex, smoking status, and area-level indicators of socioeconomic status (p > 0.2). p-Value, Cochran’s Q test for heterogeneity.