Estimating Lung Volume During Cough: A Comparison of Respiratory Calibration Tasks and Methodologies

Effective cough function requires sufficient respiratory support. To estimate lung volume, respiratory inductance plethysmography measures circumferential changes of the rib cage (RC) and abdomen (AB) during various behaviors, such as coughing. During speech breathing, the accuracy of these estimates is influenced by calibration tasks and analysis methods. Measurement error can introduce bias and confound results, yet the optimal approach for assessing lung volume during cough remains unclear.

Twenty participants with Parkinson's disease (M_age = 69 years; M disease duration = 11.39 years) completed three respiratory calibration tasks: (a) rest breathing, (b) cough-like breathing ("breathe in like you're going to cough, then breathe out forcefully without coughing"), and (c) single voluntary coughs ("cough hard one time"). Lung volume estimation error was calculated by comparing the estimated lung volume signal to the spirometry signal across tasks and task combinations. Error was also assessed across three analysis methods: the Banzett method (2:1 weighting for RC and AB) and two least squares methods-one correcting for both the RC and AB signals (LsqRC/AB) and another holding the AB constant (LsqRC).

Mean lung volume estimation error was 4.68% for LsqRC/AB, 9.88% for LsqRC, and 14.24% for the Banzett method. LsqRC/AB yielded significantly lower estimation error than both the LsqRC (p < .001, d = -1.14) and Banzett methods (p < .001, d = 1.69). Calibration task had no significant effect on estimation error (p = .889).

The least squares method correcting for both the RC and AB (LsqRC/AB) provides the most precise lung volume estimates during cough. Error associated with the Banzett method exceeded previously reported values for speech breathing by more than 50%. Additional calibration tasks beyond rest breathing may not be necessary to meaningfully reduce lung volume error for cough measurement.

https://doi.org/10.23641/asha.29657360.