I want to thank the committee of the AANS meeting for inviting us to present our work at this virtual scientific session, the Trauma Critical Care session. The work we're going to be presenting is the Electronic ICU Remote Clinician Sedation Level Verification of Nursing Assessments Corrects Sedation Discordance Between Charted and Observed RAS Scores in Mechanically Ventilated Neuro-ICU Patients. I have no disclosures to report, and neither do my co-authors, except for thanking the NREF, Neurosurgery Research and Education Foundation, for their grant that supported this work through the NREF Young Clinician Investigator Award 2019-2020. Without them, this work wouldn't have been possible, so huge thanks to them for that. The EICU is a way that clinicians from a remote site can monitor patients' labs, vitals, and their assessments, and help clinicians that are at the bedside monitor their patients and make interventions that lead to better outcomes. There's a camera and audio system in each room that they can interact and examine patients and interact and discuss patients' care with clinicians. The Richmond Agitation Sedation Scale is an ICU tool used to standardize sedation level assessment through a standard language and ranges from plus 4, where patients are combative or violent, to minus 5, where they're unarousable. As any neurosurgeon or a neuroICU clinician knows, the ordered RAS may not always reflect the actual observed bedside RAS scores. When the ordered RAS scores are not followed and patients end up more deeply sedated, this over sedation can lead to prolonged ventilator days and increased ICU length of stays, overall length of stays, and subsequent complications and worse outcomes. Many times with neuropatients that aren't in an acute ICP crisis, we want them comfortable in the ventilators, but not too sedated, and oftentimes we'll target a RAS score of zero or minus one. Therefore, we had developed a research question of, does the electronic ICU remote clinician sedation level verification of nursing assessments correct the sedation discordance that we believe would often be seen when there's a discrepancy between charted and observed RAS scores in mechanically ventilated neuroICU patients? Therefore, we designed a prospective study that would consist of two steps. The first would be to verify that indeed a RAS discordance existed between what was charted and what the actual RAS was. The way we did that was by training medical students to perform RAS assessments. They had to take a training course and do written and practical examinations, which they had to master and pass to demonstrate proficiency prior to the study. We then used an online random time generator to create times for both day and night shifts for the students to check on patients, and the inclusion criteria was mechanically ventilated neuroICU patients that were not going through an acute ICP crisis and had RAS goals entered in. The second part of the study was when we could verify that there was the RAS discordance we expected to see, we would use the EICU remote clinician as a way to help nurses have more accurate RAS concordance by using the EICU clinician to come in with them during their assessments and verify what their RAS scores were as they would do a joint neurological examination together. We believed that by having this EICU clinician perform that assessment with the nurses at random intervals that that would lead to more RAS concordance. The RAS scale ranges from minus 5 to plus 5, and the differences in RAS scoring were compared pre- and post-intervention by creating a difference variable, or a delta, to see how large the difference in RAS scoring measured between either the medical student or the EICU remote clinician assessments in combination with the nursing assessments. The data was organized by pre- or post-intervention, and the shifts between nurses were dichotomized into the day shift and night shift to visualize changes in delta and variance of the RAS scores between shifts. Chi-square tests with a p-value less than .05 for significance were used, and variation between shifts of nursing RAS scores were also taken and compared to the overall variation of RAS scores by month pre- and post-intervention. When looking at our results on the pre-intervention medical student verification side, this is showing the ranges, or the delta, where differences between what the nurses were charting and what the medical students actually observed. You can observe that there's a very wide range from 5 points off all the way to being matched or 2 points, but the big thing that sticks out is that many times their assessments are 3, 4, 5 points off, and when you compare that to the post-EICU clinician verification results, over 90% of the results once the EICU clinician was used had RAS assessments that were either matched exactly or were only 1 point off. When looking at the 0 and minus 1 part on the medical student verification, it can be seen that those numbers were less than 20%. Interestingly, and somewhat surprisingly to us, the RAS assessments between the day and night shifts revealed no difference in the RAS assessment concordance. Our initial hypothesis of over-sedation being worse during day shifts was not accurate, as there were no significant changes found. This upper left graph shows the distribution of delta medical student verification, and this shows you the wide discrepancies between what was charted and what was actually observed. The bottom graph on the left side of the pre-intervention column shows those changes between day and night shift, with the day shift being in red, night shift being in blue. These post-intervention or post-EICU assistance in verification of RAS sedation levels shows that that 90% of the scores either match or within 1 point. This next slide shows as well how this wide variance on the left pre-intervention, the delta or the discrepancy in variance between what was charted and what was observed had a wide range, but on the right, post-intervention with the ACU, the vast majority, over 90%, were between 0 and 1 point off. This slide just shows that as you look at the standard deviation pre- and post-intervention, your standard deviation was much wider prior to the EICU being used. And on the right, we see the standard deviation was much smaller post-intervention. This shows us on this slide that the patients had the similar volumes and similar injury severities so that the results, which differed so greatly, cannot be attributed to a change in the patient volume or a change in the extent of the illness of the patients. This graph shows that the actual to predicted ICU length of stay underwent a significant drop from the time we observed patients with the medical student assessments in the beginning to when we implemented the formal program with the EICU remote clinician and showed that that dropped from 1.41 to 1.14. Correspondingly, at the same time, the Apache ventilator day ratio also went from above 2 down to 1.32 over this short interval period. Another way of showing that in this table form, showing the actual and the predicted Apache ventilator days, also shows that significant drop. In conclusion, the EICU remote clinician helped eliminate significant RAS discordance between charted and actual RAS scores made by the nurses during their assessments. There were subsequently corresponding improvements seen in the ICU length of stay and the Apache ventilator day ratio as a result of the ICU intervention. We plan to continue studying prospective interventions using the EICU to help clinicians obtain more accurate sedation and better results in patients. Thank you for your time.

AANS meeting

Electronic ICU

sedation levels

mechanically ventilated neuroICU patients

prospective study