Oxygenation advisor recommends appropriate positive end expiratory pressure and FIO2 settings: retrospective validation study.

A decision support, rule-based oxygenation advisor that provides guidance for setting positive end expiratory pressure (PEEP) and fractional inhaled oxygen concentration (FIO2) for patients with respiratory failure is described. The target oxygenation goal is to achieve and maintain pulse oximeter oxygen saturation (SpO2) ≥ 88 and ≤ 95%, as posited by the Acute Respiratory Distress Syndrome Network, by recommending appropriate combinations of PEEP and FIO2. For patient safety, the oxygenation advisor monitors mean arterial blood pressure (MAP) to ensure it is ≥ 65 mmHg for hemodynamic stability and inspiratory plateau pressure (Pplt) so it is ≤ 30 cm H2O for lung protection. The purpose of this validation study was to compare attending physicians' recommendations to those recommendations of the oxygenation advisor for setting PEEP and FIO2. Adults with respiratory failure (n = 117) receiving ventilatory support were studied. PEEP, FIO2, SpO2, MAP, and Pplt are input variables into the advisor. Recommendations to increase, maintain, or decrease PEEP and FIO2 are the oxygenation advisor's output variables. Physicians' recommendations for setting PEEP and FIO2 were recorded; the oxygenation advisor's recommendations were also recorded for comparison. At all times, ventilator settings were based on recommendations from attending physicians. PEEP ranged from 2 to 22 cm H2O and FIO2 ranged from 0.30 to 0.65. A total of 326 recommendations by the oxygenation advisor and attending physicians were made to increase, maintain, or decrease PEEP and FIO2. There was a very significant relationship (p < 0.0001) between recommendations of the oxygenation advisor and attending physicians for setting PEEP and FIO2. The agreement rate for recommendations by the oxygenation advisor and attending physicians was 92%. The K statistic, a test of the strength of agreement of recommendations between the oxygenation advisor and attending physicians, was 0.82 (p < 0.0001), indicating "almost perfect agreement". Relationships for recommendations made by the oxygenation advisor and attending physicians for setting PEEP and FIO2 were excellent, PEEP: r = 0.98 (p < 0.01), r(2) = 0.96; FIO2: r = 0.91 (p < 0.01), r(2) = 0.83, bias and precision values were negligible. A novel oxygenation advisor provided continuous and automatic recommendations for setting PEEP and FIO2 that were shown to be as good as the clinical judgment of experienced attending physicians. For all patients, the target oxygenation goal was achieved. Concerning patient safety, the oxygenation advisor detected those occasions when MAP and Pplt were in potentially unsafe ranges.