A Scoring Tool That Identifies the Need for Positive-Pressure Ventilation and Determines the Effectiveness of Allocated Respiratory Therapy.

Abstract

Hospital-acquired pneumonia (HAP) and the need for positive-pressure ventilation (PPV) are significant postoperative pulmonary complications (PPCs) that increase patients' lengths of stay, mortality, and costs. Current tools used to predict PPCs use nonmodifiable preoperative factors; thus, they cannot assess provided respiratory therapy effectiveness. The Respiratory Assessment and Allocation of Therapy (RAAT) tool was created to identify HAP and the need for PPV and assist in assigning respiratory therapies. This study aimed to assess the RAAT tool's reliability and validity and determine if allocated respiratory procedures based on scores prevented HAP and the need for PPV. Electronic medical record data for nonintubated surgical ICU subjects scored with the RAAT tool were pulled from July 1, 2015-January 31, 2016, using a consecutive sampling technique. Sensitivity, specificity, and jackknife analysis were generated based on total RAAT scores. A unit-weighted analysis and mean differences of consecutive RAAT scores were analyzed with RAAT total scores ≥ 10 and the need for PPV. The first or second RAAT score of ≤ 5 (unlikely to receive PPV) and ≥ 10 (likely to receive PPV) provided a sensitivity of 0.833 and 0.783 and specificity of 0.761 and 0.804, respectively. Jackknifed sensitivity and specificity for identified cutoffs above were 0.800-0.917 and 0.775-0.739 for the first RAAT score and 0.667-0.889 and 0.815-0.79 for the second RAAT score. The initial RAAT scores of ≥ 10 predicted the need for PPV (P < .001) and was associated with higher in-hospital mortality (P < .001). Mean differences between consecutive RAAT scores revealed decreasing scores did not need PPV. The RAAT scoring tool demonstrated an association with the need for PPV using modifiable factors and appears to provide a quantitative method of determining if allocated respiratory therapy is effective.