Development of a Respiratory Support Score as a Visualization Tool in Intensive Care

In the modern era, many devices exist to support patients with respiratory insufficiency. There is currently no way to depict changes in the degree of support a patient is receiving over time.

We enrolled 4,889 subjects undergoing 5,732 cardiac surgical visits between 2011 and 2017 and extracted data elements related to respiratory support from the electronic medical record. We created an algorithm to use these data to categorize a subject's respiratory support type and to calculate an empirically derived respiratory support score (RSS) at each postoperative minute; the RSS is scored on a scale of 0 to 100. The RSS was then used to identify the timing and incidence of nonprocedural re-intubations, which were electronically verified against secondary verification fields (eg, nursing extubation note). Rates of nonprocedural re-intubations and noninvasive ventilation were compared between surgical mortality risk scores (STAT scores).

Computerized assignment of RSS was performed for 3 million subject time points. Mechanical ventilation duration varied significantly by STAT score (P < .001). Nonprocedural re-intubations increased nonsignificantly with increasing STAT score (P = .059, overall 4.3%); time to nonprocedural re-intubation did not (P = .53). Noninvasive ventilation use was more common and was prolonged with increasing STAT score (P < .001).

Elements of respiratory support can be automatically extracted and transformed into a numerical RSS for visualization of respiratory course. The RSS provides a clear visual depiction of respiratory care over time, particularly in subjects with a complex ICU course. The score also allows for the automated adjudication of meaningful end points, including timing of extubation and incidence of nonprocedural re-intubation.