Résumé :
|
Introduction : PSV is an effective treatment to relieve hypercapnic respiratory failure. However, poor patient-ventilator interactions may occur and may affect sleep efficiency and ventilation tolearnce. Our objectives were i) to identify asynchronies and leaks during sleep under PSV, and ii) to investigate the relations between asynchronies, normal ventilatory cycles and leaks during long-term recordings. Methods : 35 nocturnal ventilatory tracings (20 OHS patients and 15 COPD) were analyzed (~5000 ventilatory cycles per tracing). With a specific algorithm using flow and airway pressure, each ventilatory cycle was classified as either a nontriggered cycle (NTC), a deffective expiratory cycling (DEC), or a normal cycle (N) and described by four parameters : its maximum of pressure Pmax, its total duration Ttot, its pressurization time Tp and itsmaximumof total leak conductance G. The statistical distributionsDfor these 4 parameters were analysed andwere considered as normal when they presented a single peak : one peak at the IPAP level for DPmax, around the mean of total durations for DTtot, around the mean of inspiratory duration for DTp, and around the conductance related to the intentional leaks for DG. Using a probabilistic approach (Markov Matrix), we calculated the probability to switch fromone ventilatory cycle category to another. Results : Four profiles of patient-ventilator interactions were obtained : - profile 1 when the four distributions were normal (14 patients) : no asynchrony and no leak; - profile 1L when only DG was abnormal (5 patients), no asynchrony but occurrence of leaks; - profile 2 when only DTp was abnormal (4 patients), high rate of DEC and no leak; - profile 2L when all distributions were abnormal (18 patients), many asynchronies and major leaks. In the first two profiles, each cycle (NTC, DEC,N) had a high probability to be followed by a normal cycle. In the third profile, all cycles had a high probability to be followed by a DEC. In the last profile, a privileged relation was found between DEC and CD. Conclusion : A mathematical analysis of all individual ventilatory cycle characteristics provides important information to assess the quality of patient-ventilator interactions during sleep under PSV.
|