- The staggering number of patients requiring ventilatory support during COVID and post COVID will place significant economic pressure on the already burdened US critical care system.
- Prolonged mechanical ventilation not only increases the clinical burden of complications such as ventilator-associated lung injury, pneumonia but may also induce long-lasting laryngeal damage from prolonged intubation and further increase the cost of caring for critically ill patients.
- Strategies to optimize the early weaning of patients off ventilatory support to improve clinical outcomes and reduce mortality can be heavily supported by AI data-driven clinician support predictive tools.
This year, medical clinicians and the USA media have focused much attention on the COVID-19 pandemic causing a surge in critically ill patients requiring ventilation in the ICU. Indeed, a projected estimate back in March for the number of COVID-19 ICU patients requiring ventilatory support in the USA was a staggering 960,000. This is on top of the normal 13-20 million critically ill patients intubated in ICUs annually worldwide, of whom up to 50% require mechanical ventilation. In this article we will discuss the clinical and financial costs of prolonged intubation and mechanical ventilation; the reasons for delayed weaning; and introduce the concept of AI-driven technologies to reduce these costs and allow for earlier safe weaning in appropriate patients.
Laryngeal damage as a consequence of prolonged intubation with mechanical ventilation
Endotracheal intubation is required to protect a patient’s central airway and provide mechanical support for ventilation. Despite such a procedure being necessary to preserve life, its iatrogenic effects are often under-recognized. These include short- and long-term harmful effects that may require further medical care extending beyond the ICU period. One such harmful effect is laryngeal injury. There are no published guidelines for laryngeal assessment post-extubation, yet patients often complain of hoarseness, loss of voice, throat clearing, sore throat, and vocal fatigue post-extubation. While laryngeal injuries from intubations during surgery are mainly minor, this is certainly not the case for critically ill patients intubated in the ICU. These patients usually experience longer intubations which result in more prevalent, potentially more severe, and frequently overlooked clinical symptoms which can be responsible for voice and swallowing dysfunction. A recent systematic review of nine studies with a total of 775 patients showed a high incidence of moderate to severe laryngeal injury post-extubation.
Specifically, 83% patients presented with laryngeal injury of whom 13-31% were classified as having moderate to severe injuries.
These figures were reported for patients with a mean duration of 8.2 days on mechanical intubation in the ICU. Such moderate to severe laryngeal injuries can further increase costs due to subsequent readmissions for repair ($6000 according to 2018 figures). Laryngeal damage also has add-on negative effects such as an increased risk for dysphagia and aspiration pneumonia. Hospital acquired pneumonia can extend a patient’s ICU stay by up to 8 days, again significantly increasing required resources and costs.
Prolonged mechanical ventilation Pitfalls: VALI and VAP
Laryngeal damage is just one consequence of intubation for mechanical ventilation. Another common complication is ventilator-associated pneumonia (VAP). The risk of VAP can be somewhat mitigated by the use of higher backrest elevations and oral care strategies. Prolonged sedation is however associated with significant risks and is unavoidable. Mechanical ventilation requires sedation to provide physiological stability, maintain ventilator synchrony, and ensure patient comfort. Although there are international recommendations in place, unfortunately there are considerable discrepancies between the recommendations and actual sedation practices used in ICUs.
We see that patients who receive prolonged mechanical ventilation (PMV) have a high resource utilization, have a long ICU length of stay, and may suffer from relatively poor outcomes such as an increased risk for mortality due to ventilator-associated lung injury (VALI) and pneumonia. This holds true especially for the elderly population who are increasing in number and are at increased risk with the COVID-19 pandemic.
This begs the question as to how do we reduce the clinical and economic costs of PMV that may rapidly compromise care and are unlikely to be sustainable?
One approach is to classify patients according to their clinical needs. That is, identify patients who are likely to have unfavorable outcomes and monitor them more closely while providing a lower intensity of care and freeing up ICU beds of patients who have had a successful extubation. Another strategic approach is to enable early weaning from ventilators.
To wean, to wean early, when not to wean prematurely, these are the questions
Weaning, the progressive decline in the amount of ventilation support that a patient receives from a ventilator, accounts for approximately 40% of the total time spent on mechanical ventilation. Patients who wean successfully at the earliest opportunity have less morbidity and mortality and use fewer resources than patients who require PMV. The practice of Daily Sedation Interruption (DSI), a method to reduce the complications of prolonged sedation, has been reported to be a safe and beneficial approach to facilitating weaning. However, the best practices in the ICU setting are the daily assessment of a patient’s readiness to wean and the gradual decrease in ventilation parameters according to clinical recommendations. Some mechanical ventilation parameters, such as the fraction of inspired oxygen (FiO2) and the positive end-respiratory pressure (PEEP), are associated with lung injury if higher-than-needed values are used. This means that as long as a patient is ventilated, it is essential to use the lowest values that allow for adequate oxygenation and ventilation.
It follows that identifying patients eligible for a decrease in these values is of the utmost importance.
In fact, weaning protocols are based on a gradual decrease in both FiO2 and PEEP following a step-by-step protocol, since these parameters are used to assess a patient’s potential readiness to be weaned.
Weaning is often delayed in the ICU beyond the point where it could be started. This is frequently due to the real-life limitations of physicians’ time and resources. For instance, the fact that a patient might be ready to begin the process of weaning may only come to attention during morning rounds, since it is usually only the acutely deteriorating and critically ill patients that receive attention at other times, especially in under-resourced ICUs. This can lead to the delayed tapering-off of sedation, a longer waiting period for a patient to be alert enough for a spontaneous breathing trial (SBT), and postponing extubation until the next day to avoid risking such procedures in the middle of the night.
Protocols for weaning off of mechanical ventilation currently exist and if implemented properly can reduce the duration of weaning. However, as with all clinical protocols, they naturally rely on the compliant intervention of medical staff. This can be an issue as a 2016 Cochrane Review of 11 studies totaling 267 ICU patients found. Namely, medical staff will tend to use their own clinical knowledge, skills and judgment which may be at odds with the protocols. Furthermore, nurses will use the protocols according to their own personal confidence in their clinical decision making. Communication, training and assessing the critical care staff’s comfort and understanding of protocols, can all enhance the implementation, utilization and clinical confidence with protocols. Nevertheless, if we want to remove the human factor of compliance with protocols, automated weaning is likely to be superior to manual weaning. This is because automated weaning can rapidly recognize a deviation from a desired behavior and enforce compliance with a standardized weaning strategy which is not subject to external influences.
The AI Advantage: SMART support tools for intensivists facing complex decisions
Artificial Intelligence (AI) technologies such as CLEW are able to assist clinicians who face complex decisions by producing “smart lists” of patients that fulfill clinician-defined criteria. These can be used for the early identification of patients suitable for weaning trials. The advantage of this tool is that it is fully flexible and customizable, allowing physicians to define any new practice supporting their workflow.
While the best practice guidelines are defined by the clinicians themselves, an AI system continuously monitors patients who fulfill criteria for early weaning and displays real-time lists of patients for whom early weaning interventions may be appropriate according to those best practices.
These AI developed algorithms therefore utilize a data-driven approach to help the clinician optimize the weaning off mechanical ventilation and sedative dosing for ICU patients. The value of such an automated process that continuously monitors patients’ clinical data for weaning readiness is significant. Such technology provides real time alerts on a patient’s readiness for weaning, allowing for tapering off sedation even in the middle of the night so that by morning the patient is ready for an SBT. It follows that the clinical and economic costs of PMV can be significantly reduced.
With the COVID-19 pandemic, mechanical ventilators may be used more frequently than in previous years. Removing COVID-19 from the equation, the growing burden of an elderly population needing ICU and ventilation is also exponentially growing. By harnessing the AI-driven technologies we now have at our disposal and that are being increasingly utilized out of necessity due to the pandemic, we can ensure that patients are on ventilatory support for the minimum duration necessary and consequently improve outcomes while reducing the clinical burden and economic costs of PMV.