|| Checking for direct PDF access through Ovid
To meet some of the limitations of previous studies using grouped-analysis, we developed a cohort of persons with congestive heart failure and used a nested case-control design to estimate whether daily non-accidental mortality was associated with spatially-resolved, daily exposures to ambient NO2 and O3, and whether these associations were modified according to worsening of health.We constructed the cohort from health databases. For each case we randomly selected 100 controls matching on the calendar day and gender. We assigned daily NO2 (all year) and O3 (May-September) exposures from the following methods to predict spatially-resolved, daily concentrations at participants’ residence:measurements at the nearest monitor;inverse–distance weighting interpolation (IDW);back–extrapolation from a land–use regression model from a dense monitoring survey.We modelled air pollution using DLNMs over lags 0 to 3 days. We adjusted for age and area-based contextual variables, whereas weather and temporal trends were controlled by design. We assessed potential effect modification by some indicators possibly reflecting a worsening of health.The cohort included 65 534 individuals followed on average for four years. For NO2 and O3 all response functions were linear. For NO2, we found positive association using the back-extrapolation method (MPC: 3.0%; 95% CI: −0.8% to 7.0%) but negative associations using the nearest station (MPC: −5.5%; 95% CI: −8.0 to −2.8%), and IDW (MPC: −8.8%; 95% CI: −15.0 to −2.1%). For O3, we found positive cumulative effect for the nearest station (6.7%; 95% CI: 0.3% to 13.5%), inverse-distance weighting (18.5%; 95% CI: −2.6% to 44.1%) and back-extrapolation (7.3%; 95% CI: 3.0% to 11.9%). For both pollutants, we found evidence of effect modification according to the prescribed dose of an important drug used to treat heart failure.Exposure to ambient NO2 and O3 was associated with daily mortality in congestive heart failure. The methods used to assess daily exposure can have considerable influence on the estimated acute health effects.