Abstract Number: 439 | ID: 2017-439
Case-Crossover and Nested Case-Control Analyses of the Association between Mortality and Spatially-Resolved Daily Ambient Concentrations of NO2 among Elderly Persons with Congestive Heart Failure, Montreal, 1991-2002
Stephane Buteau(Department of Medicine, McGill University, Canada, firstname.lastname@example.org), Richard T. Burnett(Population Studies Division, Health Canada, Canada), Marie-France Valois(Department of Medicine, McGill University, Canada), Mark Goldberg(Department of Medicine, McGill University, Canada)Background/Aim: We used time-stratified case-crossover and nested case-control designs to determine the association between daily non-accidental mortality among elderly persons with congestive heart failure using different spatiotemporal prediction models of nitrogen dioxide (NO2) in Montreal, 1991-2002.
Methods: For the case-crossover analysis we matched the day in which each subject died to all similar weekdays within the same month. In the nested case-control we randomly selected 100 controls matched to each case by day of death and gender. We adjusted for age and area-based indicators of socio-economic status, whereas weather and temporal trends were controlled by design. Exposure was assigned from different methods predicting spatially-resolved, daily residential NO2 concentrations: 1) concentrations from the nearest monitor; 2) inverse-distance weighting interpolation; 3) back-extrapolation from a land-use regression model from a dense monitoring survey; and 4) the mean of all stations (case-crossover only). We accounted for the effects of air pollution and weather (case-crossover only) on mortality using distributed lag nonlinear models over lags 0 to 3 days.
Results: All response functions were linear. For the case-crossover design, results were consistent across the different exposure metrics. The mean percent change (MPC) accumulated over lags 0 to 3-days across exposure models ranged from 2.6% (95%CI: -0.6, 5.9%) to 3.4% (95%CI: 0.0, 6.9%) per increase equal to the interquartile range (i.e., approximately 10 ppb). For the nested case-control results using the back-extrapolation method, we found a MPC of 3.0% (95%CI: -0.8, 7.0%), consistent to that of the case-crossover analysis. However, we found negative associations using the nearest station approach (MPC: -5.5%; 95%CI: -8.0, -2.8%), and using inverse-distance weighting (MPC: -8.8%; 95%CI: -15.0, -2.1%).
Conclusions: Exposure to ambient NO2 may increase the risk of daily mortality in congestive heart failure, but for some exposure metrics the associations were not consistent across the two designs.