No association between major football games and cardiovascular mortality.

To the Editor: Readers of our recent paper1 expressed an interest in knowing the relative risk (RR) estimates, using the concentrations of air pollutants measured at traffic-influenced stations, rather than the concentrations measured at background stations, for the population living along busy roads in Amsterdam. In Table 1, the RR estimates obtained using traffic sites are presented together with those obtained using background sites. Compared with background stations, the relative risk estimates at traffic stations are lower for black smoke and NO2 and are not different for the other pollutants. This result is similar to that observed for the total population. The lower variance in daily air pollutant concentrations at the background stations compared with those at the traffic stations may have caused this. Analogously, a smaller range in concentrations is used to explain the same range in mortality counts. Using traffic stations produces comparable RR estimates for most compounds in the total population and in the population living along busy streets. For black smoke, we calculated the largest differences: lag 1 (RR 1.122 [95% confidence interval (CI) 1.023–1.231] vs RR 1.030 [95% CI 0.814– 1.303]) and lag 2 (RR 1.124 [95% CI 1.024–1.234] vs RR 1.286 [95% CI 1.018–1.625]). We concluded that the smaller size of the population living along busy streets (about 10% of the total population) probably results in larger confidence intervals of the risk estimates. On the other hand, we used the same range in concentrations to explain different ranges in mortality counts. A given increase in pollutant concentration is expected to result in a smaller increase in mortality counts in the population living along busy streets than in the total population. So, with these data, it remains impossible to decide whether stronger associations exist for the traffic population.