Non-invasive determination of alveolar pressure during mechanical ventilation

Abstract

The development of inadvertent positive end-expiratory pressure (PEEP(i)) in ventilated infants is of clinical relevance and difficult to measure non-invasively. A method for estimating end-expiratory alveolar pressure by applying a multiple regression analysis to airway opening pressure, flow and volume recordings during mechanical ventilation was evaluated. In eight open-chested, paralysed and mechanically ventilated mongrel dogs, alveolar pressure was measured directly with 'alveolar capsules'. Alteration of ventilation patterns and addition of a resistive element were used in three dogs to induce different levels of PEEP(i). End-expiratory alveolar pressure measured directly and determined from multilinear regression of airway opening pressure correlated well (mean error 0.06 ± 0.53 (±SD) hPa, limits of agreement -1.16 to +1.04 hPa). The other five dogs received inhalation challenges, two with histamine, two with hypertonic saline and one with methacholine resulting in a mean increase of respiratory system resistance of 230% (range 141-489%) of the baseline values. The mean error in determining PEEP(i) was 0.54 ± 0.37 hPa, the limits of agreement were -0.20 to 1.28 hPa. The method was then applied to seven mechanically-ventilated children (aged 2 months to 8 yrs, weight 4.9-23.5 kg) and the results were compared to the pressure at which inspiration began (equalling PEEP(i)). Seventy eight measurements were performed during open heart surgery, while compliance changed by between 3 and 186% of baseline values due to the surgical procedures. PEEP(i) estimated by multiple regression agreed well with the pressure at which inspiration began (mean difference 0.25 ± 0.68 hPa, limits of agreement -1.12 to 1.62 hPa). A computer model was used to determine the stability of the multiple regression method under conditions likely to stress the assumptions underlying this technique, i.e. in the presence of a high flow dependent endotracheal tube resistance and after introduction of noise in the simulated signals. Values usually encountered in clinical practice did not result in unacceptable errors in determining PEEP(i) from multiple regression analysis of airway pressure.