Thursday, January 17, 2013

FiO2 Titration Augmented by Artificial Intelligence

I have been a Registered Respiratory Therapist for greater than twelve years and have never considered the titration of oxygen a big deal. Furthermore, practicing at > 10 different facilities in eight different States I have never needed to be prompted by a physician or surveillance system to titrate the FiO2 during mechanical. Fortunately, I have only worked at facilities with respiratory care practitioner driven protocols. So when two recent articles were published on the subject within the last month I took notice [1, 2].

The first article "Titration of Inspired Oxygen Levels during Mechanical Ventilation through a Respiratory Therapist Driven Approach Based on an Electronic Surveillance System (tools). Researchers at the Mayo clinic (Minnesota) compared the weaning of oxygen during mechanical ventilation via a Respiratory Therapist encouraged by a surveillance system versus physician assessment.

The results of the study were not unexpected however; a few intricacies regarding the institution disturbed me. So what is troubling with this institution and the study itself?

Physician assessment is the standard of care
The article stated that the standard for setting the FiO2 was based on physician assessment. I can see this standard for initial settings and the very acute phase of mechanical ventilation but weaning FiO2? This is why the results are not surprising if one has to wait for physician assessment before FiO2 is weaned then at most the FiO2 would only be decreased twice per a twenty for hour period [3].  In contrast if a respiratory care driven protocol was in place the weaning of FiO2 could be performed numerous times throughout a 24 hour period, since RCPs’ are in-house 24 hours.

The investment of creating an electronic surveillance system
This facility actually invested time, money, and resources into developing an electronic surveillance system.

This poses various questions;

Why does your Respiratory Care Practitioners (RCP’s) need to be prompted to wean FiO2?
-Do the RCP’s lack critical thinking skills? Is this an educational issue?
-Are the RCP’s under-staffed? Are they out of the ICU, administering non-indicated
breathing treatments?
-Do the physicians not trust the RCP’s is this why they need to be prompted and physician assessment is standard of care?

Computer programmers are not cheap and usually make more than a RCP (unless you outsource them from India) so I believe these allocations could have been used more effectively.

That the study was actually performed
The institution actually devoted time and finances for this study when a fifth grader familiar with probability could accurately predict the results [4].

The second article [2] introduced automated closed loop titration of Fio2, provided a synopsis of operational algorithms for these systems, presented patient studies, and provided key considerations and possible disadvantages of utilizing these systems.
The key driver for the development of these systems is to prevent complications related to hyperoxia and hypoxemia in infants. The article states that sampled patients spent ~half the time under the intended SpO2 range and greater than a third of the time above the intended SpO2 range.

A key consideration presented is that “hypoxemia spells in preterm infants are largely triggered by changes in ventilation and lung volume” [2]. Increasing the FiO2 may briefly resolve the hypoxic episode, however does not address the root cause, supporting the need for an engage practitioner.

Both of these articles present using artificial intelligence systems for the titration of FiO2. These systems may augment the titration of FiO2 however these do not replace the expertise of the device operator.


[2]. Claure N, Bancalari E. Automated closed loop control of inspired oxygen concentration. Respir Care 2013; 58 (1): 151-161.


[3]. A physician usually rounds on patients in the ICU one-to-two times in a twenty four hour period and most likely not at during the night.
[4]. The probability of a RCP weaning FiO2 versus physician is much greater. A RCP is staffed every 24 hours in the ICU and ventilator patient assessments are performed at least every four hours. So a RCP team will assess the patient six (6) times within a 24 hour period versus two (2) times for a physician assessment.