Getting to Patients’ Heads Through Their Hearts*

    loading  Checking for direct PDF access through Ovid


In this issue of Critical Care Medicine, Nagaraj et al (1) describe their continued work in developing a method to predict sedation levels from heart rate variability (HRV) measurements. The work in this study is an enhancement from their prior contributions, because they demonstrate the ability to personalize the metrics generated by calibrating against prior HRV measures integrated with prior Richmond Agitation Sedation Scale (RASS) scores (2). Although incremental, this does represents an advance in the field.
The importance of this work is recognized by the dedication of many investigators to help develop improved monitoring of sedation in the critical care environment. Electronic monitoring of sedation states started with the introduction of electroencephalogram (EEG) processing and direct monitoring of brain activity over 20 years ago (3, 4). This is most notably identified by the introduction of the bispectral index monitor (BIS) by Aspect Medical Systems (Norwood, MA) in 1994 (4). Aspect processed EEG tracings using a proprietary algorithm to empirically derive a unit-less sedation metric. In practice, the device provides a single number from 1 to 100 to help the user assess the depth of anesthesia, with a recommended range of 40–60. Developed and marketed as a device to avoid anesthetic awareness, the BIS monitor, as it became to be known, developed a limited presence in the ICU despite a lack of data surrounding its validity in the critical care domain and limited evidence that it reduced awareness events in the operating room (5–9).
Although multiple technologies exist to assist in the EEG and neuronal activity processing space, the concept that one can assess sedation using HRV metrics is intriguing. HRV is known to link to neural activity, particularly when examining the frequency domain (10–12). As such, it makes sense that a correlation between sedation state and HRV parameters does indeed exist. The prior work by Nagaraj et al (1) is suggestive that such a link may be assessable and useable by clinicians.
Despite the progress demonstrated by this study, there are significant limitations to the work. First and foremost, the study used a RASS measure that was routinely collected every 2 hours. There was no attempt to isolate episodes of increased agitation, which is infrequent, but highly disruptive in the ICU patient population. The data provided from this study do not demonstrate that patients were constantly at the predicted RASS, but only that they were at the level at the time of observation.
Of additional concern is the lack of a standardized sedation protocol utilized in the conduct of this study. Patients in this study received a multitude of sedatives. This includes patients that could have been receiving dexmedetomidine or other sedative medications that impact cardiac conductivity (13). These medications may have inadvertently impacted the HRV metrics and the subsequent correlation with HRV parameters. Although the authors were able to demonstrate correlation despite the medications that were being administered, the lack of any integration of the sedation being administered greatly limits the generalizability of the authors’ conclusions, particularly when considering the rapid rise in the use of α-2 agonists in ICU sedation.
Despite these elementary concerns, when examining this study and the general concept of HRV as a sedation monitor, the greatest concern is whether this is an appropriate method to monitor sedation. The authors argue that RASS and other sedation scales are subject to inter- and intraobserver variability, and that electrocardiogram (ECG) monitoring is ubiquitous in the critical care population. They argue that it would only make sense to incorporate additional knowledge from the ECG in the management of the patient.
    loading  Loading Related Articles