Diagnosing delirium in perioperative and intensive care medicine

Purpose of review This article reviews the current literature on instruments used for screening and diagnosing delirium in perioperative and intensive care medicine. It summarizes recent findings to guide clinicians and researchers in choosing the most appropriate tools. Recent findings The incidence of delirium in hospitalized patients ranges from 5% to over 50%, depending on the population of patients studied. Failure to diagnose delirium in a timely manner is associated with serious adverse outcomes, including death and institutionalization. Valid assessment tests are needed for delirium detection, as early identification and treatment of delirium may help to prevent complications. Currently, there are more than 30 available instruments, which have been developed to assist with the screening and diagnosis of delirium. However, these tools vary greatly in sensitivity, specificity, and administration time, and their overabundance challenges the selection of specific tool as well as direct comparisons and interpretation of results across studies. Summary Overlooking or misdiagnosing delirium may result in poor patient outcomes. Familiarizing healthcare workers with the variety of delirium assessments and selecting the most appropriate tool to their needs is an important step toward improving awareness and recognition of delirium.


INTRODUCTION
Delirium is characterized as an acutely developing and fluctuating disturbance of awareness, attention, and cognition. Its prevalence is high among patients across all healthcare settings [1,2], and it is generally reported to be more frequent in the elderly, in patients with preexisting cognitive impairment, and in those admitted to the intensive care unit (ICU) [3]. While signs and symptoms of delirium are consistent, the underlying pathophysiologic mechanism is variable [4]. It may be triggered by a combination of predisposing and precipitating factors and has multiple potential causes, including acute medical illness, drug use or withdrawal, trauma, or surgery [5][6][7]. In the perioperative setting, it can manifest as postoperativae delirium (POD) after surgery and anesthesia, typically starting in the recovery room and occurring up to 5 days after surgery [8][9][10][11]. According to a new nomenclature, POD is classified as a postoperative neurocognitive disorder (NCD) [12].
Numerous epidemiologic studies report widely divergent data on the incidence of delirium, depending on the cohort of patients studied, type of surgical procedure, treatment modalities, and study methodology [13]. Delirium is an acute and transient condition; however, failure to recognize delirium can lead to serious consequences such as increased mortality, greater need for long-term care and longer hospital stays for those affected [14][15][16].
It is assumed that currently, 30-40% of all delirium cases can be prevented [17][18][19]. Therefore, early detection of delirium is crucial to mitigate the risk for adverse patient outcomes [20,21,22 & ], which may be achieved with systematic screening [23,24]. To date, over 30 unique delirium assessments have been developed to identify affected patients with a relatively small allocation of personnel and time resources [25]. Reliable delirium screening and diagnostic tools are now available for clinical routine [26][27][28].

Consensus for routine delirium diagnostics
It is assumed that 50-60% of all delirium cases remain undetected in acute hospital settings [23], due to the various subtypes (i.e., hypoactive delirium), symptom fluctuation, absence of systematic diagnostic testing, or lack of training [29]. The clinical presentation of delirium may be heterogeneous and multifaceted, and the measurement of delirium and its severity pose unique challenges [30]. Therefore, the clinical implementation of an appropriate delirium assessment instrument in daily practice must consider time constraints, level of rater expertise, test characteristics (i.e., sensitivity and specificity), and limitations of the instrument. Healthcare professionals, especially those who provide direct patient care, should be aware of delirium and its negative consequences and be familiar with reliable tools that can be used to identify patients at high risk for developing delirium [31 && ]. All recent international guidelines recommend routine screening for delirium. However, there is a lack of consistency in the recommendations. The National Institute for Health and Care Excellence guideline recommends administering a valid screening tool to any hospitalized patient with risk factors, including age >65 years, cognitive impairment, dementia, hip fracture, and severe comorbidities [32]. The American Geriatric Society's clinical practice guideline recommends the use of a valid screening tool for detecting delirium in the postoperative period in patients suspected of having delirium symptoms or exhibiting acute changes in cognition [33]. The Fifth International Perioperative Neurotoxicity Working Group recommends an objective evaluation of baseline cognition with a brief screening tool during preoperative evaluation in all patients aged >65 years and in any patient with risk factors for preexisting cognitive impairment [34]. The European Society of Anaesthesiology's guideline recommends the screening for POD in all patients starting in the recovery room and during each medical shift up to postoperative day 5 [13].
Overall, delirium instruments can be broadly divided into screening and assessment tools. Screening for delirium is the first step in the diagnostic process and is used to identify high-risk patients. Both screening and assessment instruments aim to identify a need for action (i.e., prevention and therapy) [35].

Screening tools for delirium
One of the best-known screening instruments for delirium in critically ill patients is the Intensive Care Delirium Screening Checklist (ICDSC) [36]. The instrument was created for nursing staff in the ICU, can be used without specific knowledge or training, and is particularly suitable for intubated and mechanically ventilated patients. The ICDSC is based on the DSM-IV [37] and consists of eight assessment criteria: including (1) altered level of consciousness, (2) inattention, (3) disorientation, (4) hallucination/delusion/psychosis, (5) psychomotor agitation or retardation, (6) inappropriate mood or speech/language, (7) sleep/wake cycle disturbance, and (8) symptom fluctuation. Clinical observations are made every 8 to 24 h.
The Nursing Delirium Screening Scale (Nu-DESC) [38] is based on the DSM-IV and consists of five items: (1) disorientation, (2) inappropriate behavior, (3) inappropriate communication, (4) illusions or hallucinations, and (5) psychomotor retardation. The assessment requires three observations within 24 h. In a recent systematic review and meta-analysis, the NuDESC demonstrated higher sensitivity than Confusion Assessment Method (CAM) or Delirium Detection Score, while all showed high specificity (0.90 or greater) [39 & ]. However, the NuDESC does not include an interview or a cognitive test like other delirium assessment tools, which may be a disadvantage [39 & ]. In addition, the level of sedation

KEY POINTS
Delirium is a common, serious, and often preventable complication affecting older adults.
Overlooking or misdiagnosing delirium may result in poor patient outcomes.
Appropriate delirium screening and diagnostics is recommended by recent international guidelines and remains a critical area of need to ensure optimal prevention and treatment of delirium.
To date, over 30 unique delirium assessments have been developed to identify affected patients in clinical routine and research. in patients should be recorded, as this can influence the test results [27]. The Neelon and Champagne scale [40] was developed for nursing staff on general hospital wards. It was validated for use in the ICU [41], but cannot be used in intubated patients [42 & ]. It contains nine items, divided into three subscales (processing, behavior, and physiological control). While it allows grading of the severity of delirium, it is not based on international classification systems.
The Delirium Observation Screening Scale (DOSS) [43] is a screening scale for use in general nursing care. It assesses verbal and nonverbal behavior based on DSM-IV criteria. The original DOSS consists of 25 items [43], and the abbreviated 13-item DOSS form has been described by Wong et al. [44].
The 4 ''A'' test (4-AT) [45] can be used for the initial assessment of delirium and other unspecified cognitive impairments. The instrument is based on the DSM-5 and consists of four criteria: (1) alertness, (2) orientation (using the 4-item Abbreviated Mental Test) [46], (3) attention (using the Months Backward Test (MBT)) [47,48], and (4) acute change or fluctuating course. The assessment of alertness, orientation and attention is based exclusively on patient observation at the time of the examination. The assessment of the fluctuation in symptoms requires information from other sources (e.g., the medical history, information from medical staff, physician's letters, and hospital course documentation). In a recent multicenter study, the 4-AT has shown better sensitivity compared to the CAM [49]. Healthcare personnel can carry out the 4-AT without special training.
A comparison of the various characteristics of screening instruments for delirium is shown in Table 1.

Diagnostic tools for delirium
The gold standard for diagnosing delirium is based on criteria of the International Statistical Classification of Diseases and Related Health Problems (ICD-10) [50] and the Diagnostic and Statistical Manual of Mental Disorders (DSM-5) [51,52]. Compared to screening instruments, an assessment test, which allows a definitive diagnosis of delirium, is usually more time-consuming and often requires specific training of the examiner [35].
The CAM [53] is one of the most widely used instruments in delirium diagnostics because of its accuracy and ease of use. The CAM is based on DSM-III-R criteria [54] and was originally developed for physicians with no training in psychiatry. Using this instrument, delirium can be identified quickly, reliably, and in a standardized manner, both in the clinical field and for research purposes. A long (9 criteria) and short form (4 criteria) of CAM are available. The long form assesses: (1) acute onset, (2) inattention, (3) disorganized thinking, (4) altered level of consciousness, (5) disorientation, (6) memory impairment, (7) perceptual disturbance, (8) psychomotor agitation or retardation, and (9) altered sleepwake cycle. It also requires a structured approach and assesses patients based on a conversation with the examiner and brief formal cognitive testing. The short form is based on the four main characteristics of delirium: (1) acute onset and fluctuating course, (2) inattention, and either (3) [58], and the emergency department (ED) [59].
The CAM for the ICU (CAM-ICU) [56] is a modified CAM form intended for use in intubated and mechanically ventilated patients [60]. The original diagnostic algorithm was slightly adapted. The (1) acute change or fluctuating course are assessed by observation. (2) Inattention is assessed with a short test: a sequence of letters (e.g., SAVEAHAART, CASA-BLANCA) is spelled out and the patient is encouraged to squeeze the examiner's hand when saying each letter ''A''. Up to two errors are allowed. (3) Altered level of consciousness is determined using the Richmond Agitation and Sedation Scale [61]. The presence of (4) disorganized thinking is evaluated using a comprehension test [62]: four easy-toanswer yes/no questions are asked (e.g., ''Will a stone float on water?''), which can be answered with a nod of the head, handshake, or wink. The patient is also asked to hold up as many fingers as shown; up to one mistake is allowed here. Delirium is diagnosed if criteria 1, 2, and either 3 or 4 are positive.
The bCAM [57] is a modified CAM-ICU and was designed to improve sensitivity in noncritically ill patients. When performed by a physician, the bCAM has a higher sensitivity and slightly lower specificity in comparison with nonphysician examiners (sensitivity: 84% vs. 78%, specificity: 96% vs. 97%, respectively) for delirium in older patients in the ED [51]. The bCAM's diagnostic performance appears to be similar in older patients who are admitted to the hospital [63]. Although bCAM and CAM-ICU are very similar, there are two main differences. The bCAM uses the abbreviated form of the MBT [47,48] (i.e., December to July) to identify inattention. Disorganized thinking results from any mistake in the bCAM.
The 3-min Diagnostic Interview for CAM-defined Delirium (3D-CAM) [58] allows for a quick assessment of delirious states within a timeframe of ].
Tools to assess the severity of delirium The assessment tools described above have, without exception, dichotomous results (delirium yes or no), but do not indicate the severity of delirium. To date, there are a handful of instruments available for this purpose.
The CAM-Severity (CAM-S) [65] is only a means of quantifying the intensity of delirium symptoms and must be used in addition to the original CAM. The instrument is available in a long and short form, and the CAM-S short form is intended for use with the short-form CAM [65].
The Delirium Rating Scale Revised 98 (DRS-R-98) [66], which is based on DSM-IV, is used to assess the presence and severity of delirium simultaneously. It is a revised form of the original DRS from 1988 [67] and consists of 16 items. Three items (temporal onset, fluctuation of symptom severity, and physical disorders) are used to establish the diagnosis, and 13 items are used to determine the degree of severity. All available sources of information are used to assess the individual items, including family members, visitors, hospital staff, physicians, and medical records. The instrument is suitable for both the initial assessment and for repeated measurements of delirium severity. However, the DRS-R-98 may be too time-consuming (20-30 min) for clinical routine and is therefore mainly used in clinical research [65].
The Memorial Delirium Assessment Scale (MDAS) [68] was also developed to diagnose delirium and to classify its severity accordingly. This ten-item instrument integrates behavioral observations and objective cognitive tests. The MDAS does not record acute changes or a fluctuating course. The MDAS is based on DSM-IV and earlier or alternative diagnostic tools (e.g., DSM-III, DSM III-R, ICD-9).
A comprehensive summary of instruments used for the diagnosis and severity assessment of delirium is shown in Table 2.

Tools to assess preexisting cognitive symptoms
The American Society of Anesthesiologists Brain Health Initiative Summit (2018) and Global Council on Brain Health (2020) recommend the preoperative evaluation of cognitive function in older patients as a leading risk factor for the development of POD [69 && ]. However, this practice has not been widely implemented [70,71]. Different tools are recommended for initial vs. secondary screening, prioritizing higher sensitivity tools for the former and higher specificity tools for the latter [72 & ]. In addition, critical parameters such as cognitive domain, sensitivity, specificity, and time efficiency should be considered when selecting tools to screen for mild cognitive impairment (MCI) [72 & ]. Some authors suggest the use of a combination of two self-administered highly sensitive tools, the Dementia Screening Interview (AD8) [73] and the Informant Questionnaire on Cognitive Decline in the Elderly (IQCODE) [74] in initial screening, as well as a combination of two highly specific widely covered tools, the DemTect test [75] and the Montreal Cognitive Assessment (MoCA) [76], or the Memory and Executive Screening (MES) [77] and the MoCA in secondary screening [72 & ].
The IQCODE is a self-administered questionnaire assessing short-and long-term memory, orientation to time and place, financial awareness and learning skills, as well as executive functions [78]. The score has no relationship with an individual's level of education or premorbid level of intelligence. This contrasts with the Mini-Mental State Examination (MMSE) [79], which is the most used cognitive screening tool by both general practitioners and specialists in Western countries [80].
The AD8 is a self-administered screening tool that briefly assesses memory, orientation, and judgment [81]. It serves as a screening tool for cognitive status in community and primary care settings, as well as in the ED, as the AD8 is typically completed within 3 min [82]. However, it has been reported, that a combination of AD8 and MMSE is more sensitive in detecting cognitive impairment than AD8 alone [83].
The DemTect [75] includes five tasks on the following cognitive functions: short-term memory, working memory, language, number processing, and executive functions. The test is carried out in an oral and written survey within 8 to 10 min.
The MoCA seems to be the most appropriate to detect MCI, given its superior sensitivity compared to the MMSE (sensitivity: 89% vs. 62%, specificity: and 75% vs. 87%, respectively) [76,79,84 & ,85]. The MoCA tests multiple cognitive domains, including short-term memory, visuospatial skills, executive function, attention, working memory, language, and orientation. However, it is considered a relatively comprehensive test with limitations due to a relatively long administration time (10 min) and the influence of education level [86,87].
The MSE consists of two subtests: MES-5R that tests short-term memory and learning, and MES-EX that tests for executive function. The tool is independent of education level, and both subtests can be administered within 7 min [77].
The MiniCog is a short instrument, which has been validated to screen for cognitive impairment in a broad range of patients [88]. The test consists of a 3-item recall test for memory and a clock drawing test for visuospatial representation and executive function. The MiniCog is scored out of 5 maximum points, and a score of less than 3 points has been validated for dementia screening [88,89,90 & ]. The instrument has been validated as a screening tool for dementia [91] and cognitive impairment [92]. Furthermore, it was found to be feasible during the preoperative clinical evaluation and can be administered in just 3 min [93]. However, the instrument has a minimal education and language/ethnic bias [94].
In recent years, computer-aided neurocognitive test batteries have been developed to objectify cognitive performance. Computer-based test batteries have many advantages compared to paper and pencil-based test batteries (e.g., elimination of examiner-dependent differences, direct availability of the results without manual evaluation, etc.).
The digital Clock Drawing Test (dCDT) [95] is administered using a digital pen. It involves multiple cognitive domains, including visuospatial construction, executive function, and attention [96][97][98], making it a sensitive screening tool for cognitive impairment. In addition, the dCDT objectively obtains additional drawing metrics, such as the time taken to draw individual elements of the clock, a feature which offers an advantage over the standard paper-based test [95]. Furthermore, the dCDT has been proved to be acceptable to detect cognitive impairments associated with many neuropsychiatric disorders [99,100].
The CogCheck is an application in which patients independently carry out cognitive tasks on a tablet computer. It aims to assess the individual risk of postoperative NCD (e.g., delirium) in surgical patients and uses language-free cognitive tests to record temporal orientation, episodic memory, attention span, and executive function [101].
All computer-based assessments tests are limited since they only allow a rough assessment of cognitive deficits. They are not a substitute for detailed diagnostics. A comprehensive neuropsychological examination should be carried out to confirm any abnormal measurement.

Implications for clinical practice
Several systematic reviews have attempted to structure the overabundance of available screening tools  [25]. In the ICU, 95% of institutions report using the ICDSC or the CAM-ICU [103 & ]. However, a previous diagnostic meta-analysis examined and compared the accuracy of the CAM-ICU and the ICDSC against reference standards. It was identified that the CAM-ICU had higher pooled specificity compared to ICDSC (95% vs. 87%, P ¼ 0.04) [104]. Furthermore, the increasingly popular 4-AT and several CAM-derived assessments can be used in different acute care settings [55 & ]. Instruments that only rely on clinical observations tend to be more subjective. This allows more bias, and thus may compromise interrater reliability. However, the instruments can help identify certain delirium symptoms, assess the clinical course of delirium, and measure the effectiveness of various therapeutic interventions. Nevertheless, it is somewhat unlikely that a universal delirium screening tool will be discovered and disseminated soon. The choice of any assessment tool should be driven by the clinical setting, as well as staff and time resources, training, feasibility, and validity. There have been recent harmonization efforts to allow the comparison of equivalent delirium rates between studies and enable the pooling of data, regardless of the delirium identification measure used. In the future, this could help facilitate clinical decision making and multicenter research.