Impact of correcting peak flow for nonlinear errors on air pollutant effect estimates from a panel study.

Air pollutant effects are commonly investigated using panel studies employing daily measurement of changes in peak expiratory flow (PEF). Variable orifice PEF meters are inaccurate with a nonlinear relationship to actual PEF. The impact on a panel study of correcting these errors was examined. Twice-daily PEF readings were taken by 147 9-yr old children for 8 weeks and corrected using an equation derived from the response of 32 Vitalograph meters to a servomechanism-controlled pump. Pollutant effect estimates for corrected and uncorrected readings were derived using a regression approach incorporating appropriate confounders. Correction impacted little on mean PEF values (333.1-334.2 L x min(-1)), but did alter effect sizes. Nonsignificant nitrogen dioxide estimates for the entire panel decreased by up to 73%, but, for symptomatic/atopic children, a significant 5-day mean result was lost (decrease in effect size from -2.53 to -0.90% per 10 parts per billion (ppb)) and lag 0 became significant (decrease from -0.51 to -1.22% per 10 ppb). Mass concentration estimates of particles with a 50% cut-off aerodynamic diameter of 2.5 microm moved in both directions (-0.22 changed to 0.11% per 10 microg x m(-1) lag 3 and -0.29 to -0.73% per 10 microg x m(-3) for the 5-day mean). Correction of nonlinearity of peak expiratory flow meters influenced the overall outcome of this panel study, and the changes in effect estimates would be sufficient to alter the interpretation of some studies. For adults, a greater change in effect estimates may follow the larger correction required for their usual peak expiratory flow range.