Using Smart EHR Tools for Early Detection of Delirium in Hospitalized Older Adults

Abstract

<h3>Introduction</h3> Delirium is a serious medical condition associated with increased mortality, increased length of hospital stay, high rates of rehospitalization and increased hospital costs. This illness is especially common in hospitalized older adults with estimated prevalence between 10-50% of this population. Guideline based treatment of delirium consists largely of delirium prevention and management of symptoms once diagnosed. Therefore, early detection and identification of patients at high risk of developing delirium are vital strategies. In a retrospective chart review assessing the current method for detecting delirium in a large, quaternary, academic medical center, prevalence was estimated at <3% of hospitalized patients. This suggests many cases go unreported. Previous studies have shown chart based methods to be valid tools for detecting delirium in hospitalized patients. Our team developed vACE NOVAD (Virtual Acute Care for Elders, Neurological Orientation Verbal Response Assessment of Delirium): a health care analytic tool utilizing nursing assessments documented in the EHR (electronic health record) to detect delirium in hospitalized patients. The following study aimed to validate this tool by comparing it to an existing validated delirium assessment: bCAM (Brief Confusion Assessment Method). <h3>Methods</h3> Researchers used a daily automated report to identify hospitalized patients 70 and older at a large quaternary academic medical hospital in Phoenix, Arizona. Participants were chosen randomly from this report. Non-English speaking patients were excluded. Trained psychiatry resident physicians communicated with the patient's nurse and asked for the bCAM to be performed on the patient. The nurse then reported results of the assessment back to the researching physician who would score and document the results. Patients were considered to have delirium if the bCAM completed with the patient's nurse was positive. This was then compared to nurse's assessments, documented in the EHR, of patient's neurologic symptoms, best verbal response from the GCS (Glasgow Coma Scale), and orientation over the previous 72 hours: NOVAD. If a patient was documented as anything other than having no neurologic symptoms, a best verbal response on GCS of "oriented," and an orientation assessment of "oriented x4," the patient was positive for delirium based on NOVAD. <h3>Results</h3> 30 hospitalized patients over the age of 70 were included in this initial prospective validation study. 30% (n=9) of patients in this study had delirium based on bCAM. Delirium prevalence based on NOVAD score was 70% (n=30). The sensitivity and specificity of NOVAD were found to be 78% and 33% respectively. <h3>Conclusions</h3> This chart-based, health care analytic tool is an effective and timely way to aid in early identification of delirium in hospitalized older adults. Given low specificity, it is not an alternative to a formal diagnostic evaluation by a physician. Further validation of the tool with a larger participant volume is ongoing. <h3>Funding</h3> No funding to declare