The Effects of Mechanical Ventilation on the Development of Acute Respiratory Distress Syndrome

Acute Respiratory Distress Syndrome (ARDS) is a severe lung illness characterized by inflammation and fluid accumulation in the respiratory system. Historically, ARDS and other forms of respiratory failure have been treated using mechanical ventilation to help maintain gas exchange in the lungs. However, clinical investigators are beginning to discover the adverse effects of mechanical ventilation if it is not applied properly. Specifically, excessive ventilator volumes and pressures may exacerbate existing lung injury and increase hospital mortality. Furthermore, aggressive ventilation may cause lung injury and trigger an inflammatory response that is characteristic of ARDS. These findings have alarmed the critical care community, and many studies have been conducted to find mechanical ventilator settings that reduce mortality in patients with ARDS. However, there have been no firm recommendations on the optimal settings for patients who require ventilator therapy for reasons apart from respiratory failure. In this thesis, we retrospectively examine a large medical database (MIMIC-II) to study the relationship between mechanical ventilation and the development of ARDS. Specifically, our goals are to (1) find patients who did not have ARDS at the beginning of mechanical ventilation but who later developed the disease; (2) identify physiologic and ventilator-associated risk factors for ARDS; and (3) develop a text analysis algorithm to automatically extract clinical findings from radiology (chest x-ray) reports. Our findings suggest that acute respiratory distress syndrome is a relatively common illness in patients who require mechanical ventilation in the ICU (152 of 789 without ARDS at the outset eventually developed the disease). High plateau pressure (odds ratio 1.5 per 6.3 cmH20, p < 0.001) is the most important ventilator-associated risk factor for the development of new ARDS. Physiologic risk factors include high weight, low blood pH, high lactate, pneumonia, and sepsis. Thus it may be possible to reduce the occurrence of ventilator-induced lung injuries with careful pressure management. However, a randomized prospective study is needed to support this hypothesis. Thesis Supervisor: Roger Mark Title: Distinguished Professor in Health Science and Technology