Categorization in Mechanically Ventilated Pediatric Subjects: A Proposed Method to Improve Quality

Abstract

BACKGROUND: Thousands of children require mechanical ventilation each year. Although mechanical ventilation is lifesaving, it is also associated with adverse events if not properly managed. The systematic implementation of evidence-based practice through the use of guidelines and protocols has been shown to mitigate risk, yet variation in care remains prevalent. Advances in health-care technology provided the ability to stream data about mechanical ventilation and therapeutic response. Through these advances, a computer system was developed to enable the coupling of physiologic and ventilation data for real-time interpretation. Our aim was to assess the feasibility and utility of a newly developed patient categorization and scoring system to objectively measure compliance with standards of care. METHODS: We retrospectively categorized the ventilation and oxygenation statuses of subjects within our pediatric ICU utilizing 15 rules-based algorithms. Targets were predetermined based on generally accepted practices. All patient categories were calculated and presented as a percent score (0 -100%) of acceptable ventilation, acceptable oxygenation, barotrauma-free, and volutrauma-free states. RESULTS: Two hundred twenty-two subjects were identified and analyzed encompassing 1,578 d of mechanical ventilation. Median age was 3 y, median ideal body weight was 14.7 kg, and 63% were male. The median acceptable ventilation score was 84.6%, and the median acceptable oxygenation score was 70.1% (100% being maximally acceptable). Potential for ventilator-induced lung injury was broken into 2 components: barotrauma and volutrauma. There was very little potential for barotrauma, with a median barotrauma-free state of 100%. Median potential for a volutrauma-free state was 56.1%. CONCLUSIONS: We demonstrate the first patient categorization system utilizing a coordinated data-banking system and analytics to determine patient status and a surveillance of mechanical ventilation quality. Further research is needed to determine whether interventions such as visual display of variance from goal and patient categorization summaries can improve outcomes. (ClinicalTrials.gov registration NCT02184208.)

The Quality of Quality Metrics

“…We might wonder, therefore, why we have such high expectations 1 and yet so few metrics of quality performance in those areas. The study by Walsh et al 2 could be a landmark contribution in that direction. On the one hand, replication of their work will be a challenge for most institutions; what they have done involved special hardware (to port data from ventilators to the electronic health record) and custom web-based analytics software developed by their institution.…”

“…Having attempted more modest projects that involve custom software and electronic health record integration, 3 I can appreciate the formidable barriers to entry. On the other hand, Walsh et al 2 have provided a detailed set of rules that others can use to assemble quality metrics for the domains of mechanical ventilation, oxygenation, and ventilator-induced lung injury. Their definitions for rules-based algorithms might inform tools already developed in the emerging field of cognitive computing (eg, WatsonPaths).…”

“…Volutrauma is as an ultrastructural lung injury due to overdistention occurring during mechanical ventilation. 2 Barotrauma and volutrauma reflect 2 sides of the same phenomenon: lung injury due to excessive distending volumes and/or excessive airway pressure. Alveolar pressure can be easily estimated by measuring plateau pressure.…”

Daily Goals Formulation and Enhanced Visualization of Mechanical Ventilation Variance Improves Mechanical Ventilation Score

“…Subject Categorization. Subject categorization was conducted as described previously by Walsh et al 37 In brief, subject categorization uses a system of rules-based IF-THEN algorithms covering 4 clinical domains of outcomes: (1) ventilation and (2) oxygenation and ventilator-induced lung injury (VILI) described as either (3) low risk PIP or (4) low risk V T . Table 1 provides the definitions of the rules-based algorithms by acceptable category.…”

Exploring the clinical relevance of vital signs statistical calculations from a new-generation clinical information system

“…The main objective of this new approach is to apply big-data principles (volume, velocity, variety, veracity, and data value) to the concept of personalised medicine. In this sense, information has already been published about the application of high-density data CIS (HDDCIS) in intensive care contexts, both in development studies 2,3,13,[15][16][17] and as original research 4,8,12,[18][19][20][21][22][23] . Although the use of HDDCIS is not yet common in ICUs, in the future they will help revolutionise the monitoring of vital signs as we currently know it.…”