Air pollution and ST-segment depression in elderly subjects

Environmental Health Perspectives, July, 2005 by Diane R. Gold, Augusto A. Litonjua, Antonella Zanobetti, Brent A. Coull, Joel Schwartz, Gail MacCallum, Richard L. Verrier, Bruce D. Nearing, Marina J. Canner, Helen Suh, Peter H. Stone

In addition to analyses evaluating ST-segment level as a continuous outcome, we analyzed the binary response "ST-segment depression [greater than or equal to] 0.5 mm," defined as a mean ST-segment level for a given portion of the protocol of at least -0.5 mm (i.e., mean ST-segment level [less than or equal to] -0.5 mm compared with ST-segment level > -0.5 mm). This definition differed from that of classic ischemia in that it did not require within-test or within-portion of the protocol reversibility. For this secondary analysis, we fit a logistic regression model with random intercepts to data from those subjects having at least one response of each type (depressed and nondepressed ST-segment) during that particular protocol (13 of 28 study participants contributed data to at least one portion of the protocol).

Twenty-four study participants with 269 observations were included in analyses either with continuous or with binary (dichotomous) ST-segment outcomes. We had sufficient observations to evaluate the effects of between-test increases in pollution levels on between-test depression in the mean ST-level for each portion of the protocol. However, we were unable to assess the effect of between-test changes in pollution on the risk of within-test reversible ST-segment depression that fit criteria for ischemia because of the rarity and lack of variability of such events. During the study, only 5 of 28 study participants had ischemic ECG events (defined above as within-test reversible horizontal or down-sloping ST-segment depression [greater than or equal to] 0.5 mm).

Each regression model included an indicator variable for each subject, pollutant concentration, a cubic effect of the mean of the current hour temperature, and a linear trend of time. Other confounders considered included day of week and time of day, which were both highly correlated with the subject indicator variables and were thus dropped from the model. Separate models were fit using lags of 1-24 hr, as well as previous 12 and 24 hr moving averages, of pollution concentration. Finally, models containing multiple pollutant concentration as predictors were fit to account for confounding due to moderate to high correlations among different pollutant concentrations. Multiple lags and moving averages were evaluated to select the best lag structure for temperature and each individual pollutant, and models reflect these evaluations. All statistical analyses were performed using the SAS statistical software package (SAS Institute Inc., Cary, NC). The conditional linear mixed models were fit using PROC MIXED, whereas the logistic mixed models were fit using PROC NLMIXED (SAS Institute Inc.).

Estimates of the effects of BC were scaled to the difference between the 10th and the 90th percentile in levels for the appropriate lag or mean value of BC.

Results

The median age of the population was 73, and many participants had cardiac risk factors (e.g., history of hypertension, prior smoking) or coronary artery disease (Table 1). As expected, mean heart rate rose during exercise and returned to baseline at rest (Table 2) during the 269 tests for the 24 participants included in analyses. Simultaneously, median ST-segment level was lower during and immediately after exercise than at first rest. ST-segment depression was rare in the modified aVF lead, and all subsequent analyses are based on findings in the modified V5 lead, the lead that most consistently identifies myocardial ischemia when it is present (Lanza et al. 1994). Air pollution levels were only modestly elevated, and maximum levels for U.S. EPA criteria pollutants were all below accepted or proposed National Air Quality Standards (Table 3). CO levels never exceeded 2 ppm. BC levels rose early in the morning and were at their peak between 0600 and 0900 hr.