Impact of a Respiratory Therapy Assess-and-Treat Protocol on Adult Cardiothoracic ICU Readmissions.

BACKGROUND: The purpose of this retrospective medical record review was to report on recidivism to the ICU among adult postoperative cardiac and thoracic patients managed with a respiratory therapy assess-and-treat (RTAT) protocol. Our primary null hypothesis was that there would be no difference in all-cause unexpected readmissions and escalations between the RTAT group and the physician-ordered respiratory care group. Our secondary null hypothesis was that there would be no difference in primary respiratory-related readmissions, ICU length of stay, or hospital length of stay. METHODS: We reviewed 1,400 medical records of cardiac and thoracic postoperative subjects between January 2015 and October 2016. The RTAT is driven by a standardized patient assessment tool, which is completed by a registered respiratory therapist. The tool develops a respiratory severity score for each patient and directs interventions for bronchial hygiene, aerosol therapy, and lung inflation therapy based on an algorithm. The protocol period commenced on December 1, 2015, and continued through October 2016. Data relative to unplanned admissions to the ICU for all causes as well as respiratory-related causes were evaluated. RESULTS: There was a statistically significant difference in the all-cause unplanned ICU admission rate between the RTAT (5.8% [95% CI 4.3-7.9]) and the physician-ordered respiratory care (8.8% [95% CI 6.9-11.1]) groups (P = .034). There was no statistically significant difference in respiratory-related unplanned ICU admissions with RTAT (36% [95% CI 22.7-51.6]) compared with the physician-ordered respiratory care (53% [95% CI 41.1-64.8]) group (P = .09). The RTAT protocol group spent 1 d less in the ICU (P < .001) and in the hospital (P < .001). CONCLUSIONS: RTAT protocol implementation demonstrated a statistically significant reduction in all-cause ICU readmissions. The reduction in respiratory-related ICU readmissions did not reach statistical significance.