Currently physicians and nurses rely on their own clinical skills and experience to diagnose and record 'delirium' in the Electronic Health Records (EHR). This study aims to determine how delirium as a diagnosis is documented by clinicians in the EHR at Hadassah Hospital. The knowledge gained from this study will support the design of a better surveillance approach to monitoring delirium events in postoperative patients using electronic healthcare recorded data.

There is considerable uncertainty surrounding the quality of 'delirium' records in the Electronic Health Records (EHR). The reliability of this chart estimation has become questionable in the absence of an objective definition of 'delirium' and a lack of highly accurate diagnostic tools in the hospital setting.

Given the difficulty of accurately identifying delirium and the deficiency in the quality of EHR documentation, it is not surprising that delirium is grossly underestimated, undertreated, not properly recorded in the EHR or misreported. Data concordance plays a major role in documentation quality, especially for data-mining and knowledge extraction analysis, and therefore it is essential to address the reliability of 'delirium' labeled data within the EHR system.

Clinical Background::

Postoperative encephalopathy known as "delirium" is a deleterious, potentially risky, and often preventable complication representing a serious "brain failure" condition, commonly seen in the intensive care unit (ICU) setting.

The etiology of postoperative delirium (POD) is likely to be a consequence of the physiological and biochemical derangement induced by the underlying surgical pathology, surgical trauma, pain, analgesia, and anesthesia. Common pathophysiological factors, processes, and conditions leading to delirium are multifactorial and could involve neurotransmitters, inflammatory processes, drugs, and oxygenation impairment. Multifactorial perioperative factors, such as anesthesia and adjuvant drugs, also play important roles in contributing to postoperative delirium. Perioperative care is, therefore, a potential focus for investigation of data-driven evidence to understand the POD prior to developing effective intervention tool.

Delirium Burden::

Delirium creates distinct emotional distress in patients, family and caregivers.

The burden of illness due to delirium is significant, with a higher incidence of postoperative complications, prolonged length of ICU and hospital stay, resulting in 30-day mortality and unplanned readmission when compared to patients without delirium. The long-term prospects of delirious patients show poor quality of life ('QoL') indicators related to functional decline, new institutionalization, persistent cognitive impairments and higher mortality, with persistent cognitive impairments in 12% of previously "cognitively well" patients, and an even higher percentage in elderly, obese, and previously admitted patients. The postoperative delirium is associated with longer-term cognitive decline and potentially 'portends descent to dementia' .

The mortality rate associated with delirium is approximately 40% , as high as acute myocardial infarction. The total cost estimates attributable to delirium ranged from $16,303 to $64,421 per patient, implying that the national burden of delirium on the health care system ranges from $38 billion to $152 billion each year in US. This cost is highly comparable to the substantial costs of falls and diabetes, which emphasizes the need to address this costly disorder with increased timeliness and urgency.

Standard of Care::

Efforts to detect delirium have relied upon two major methods , both of which fall short of the practical needs of a modern hospital environment. Screening instruments, largely based upon chart review, well-investigated risk factors and patient interview, have been unsuccessful due to the challenges of implementing these into clinical workflows and providing ongoing training for healthcare providers to use such instruments. In addition, they exhibit poor sensitivity in routine use. While the early detection of delirium provides clear and significant advantages in effective treatment, the screening tools available for the disorder are not efficient or effective enough to do so. Delirium is often underrecognized and misdiagnosed, exemplified in a 2014 study that found that the successful detection of the disorder by staff was only at 23%, even after extensive multimodal education about the disorder and how best to detect it.

Despite its importance for patient safety and public health, delirium is often unrecognized by clinicians, therefore the effective strategies of intervention remain elusive. Moreover, the presentation of delirium is heterogeneous and multifaceted, and measurement of delirium and its severity pose unique challenges.

Unmet Need::

The need exists for an objective, affordable and reliable modern assessment with the ability of early recognition, improved screening, and continuous monitoring of postoperative encephalopathy. The potential impact of such a tool has been recognized as highly important for the establishment of evidence-based data for better tracking prognosis, monitoring response to treatment and estimating the burden of care both during and after hospitalization. Nurses themselves recognize this unmet need. Many have argued that the current screening tests are too subjective and rely on each tester's interpretation.

Concept Creation::

The era of "Medical Big Data" and next-generation health analytics is well upon us. Conceptually, "Big Data" may include data-driven clinical features considered to be hidden, uncertain, unrecoverable and unmanageable for human interpretation without the help of computerized data processing and advance data-driven algorithms. As delirium in post-surgical patients is often hard to recognize and remains a largely untreated condition, it is hypothesized that an exploratory analysis of historical medical records by using an advanced algorithm could reveal novel and improved knowledge about the nature of delirium. However, the quality, computability, reliability, accuracy and completeness of the data are questionable.

Therefore this study aims to perform a retrospective exploratory analysis of historical records locked in one or more clinical databases (i.e. Metavision, EHR, AIMS, etc.), and/or in one or more hospital settings.

Inclusion Criteria:

1. All postoperative patients aged 18 years and older; and
2. Undergoing unplanned admission to an intensive care unit (ICU); or
3. Major complications associated with either operating room procedure or anesthesia. Example: Cardiac or circulatory event and/or Cardiac arrest during or within 24 hours of operation or administration of anesthesia; or Acute myocardial infarction (AMI) during or within 48 hours of operation or administration of anesthesia; or
4. Postoperative Central Nervous System (CNS) event (e.g., CVA, seizures, coma) during or within 48 hours of operation or administration of anesthesia; AND
5. Respiratory failure not present prior to the 25th hour of hospitalization or not present before surgery;

Exclusion Criteria:

1. Delirium present on admission; or
2. any psychotic or degenerative related diagnosis on admission (e.g. Senile and presentile dementias such as Alzheimer's or Pick's dementia, Creutzfeldt-Jakob disease, Huntington's chorea, Wilson's disease).

Sample Size: 50,000 Gender: M - 25,000; F - 25000 approx. Age: 18 and older General Health Status: All postoperative patients undergoing unplanned admission to an intensive care unit (ICU).

Total Number of Subjects Projected: It is expected that approximately 75,000 subjects will be enrolled (identified for further review) to produce 50,000 evaluable subjects.