Comparison of delirium detection tools in acute care

For a tailored use of delirium screening instruments, it is important to understand the different aspects of delirium as a psychopathological syndrome and its diagnostic criteria. In the International Classification of Diseases 10th edition (ICD-10) the syndrome of delirium is described as a rather variable combination of different psychopathological disturbances with a non-specific organic etiology (Table 1; [53]). Besides altered consciousness, an alteration of attention, perception, (logical) thinking, memory, psychomotor behavior, emotion or the sleep-wake schedule might be present.

Different subtypes of delirium exist: hyperactive delirium accounts for 20% of cases and is characterized by psychomotor agitation and restlessness [43]. It often includes pulling out tubes and is disrupting for hospital procedures, and is therefore an eye-catcher diagnosis for hospital professionals. In contrast hypoactive delirium, accounting for 30% of cases, presents with reduced psychomotor activity and patients are often misdiagnosed with depression or severe dementia. About 45% of delirious patients can show both hyperactive and hypoactive symptoms indicating a fluctuating course of disease.

Delirium can also be accompanied by psychotic symptoms. More than 40% of patients display predominantly optical hallucinations or delusions, such as being poisoned or persecuted [45].

The reference standard for delirium diagnosis has most commonly been the psychopathological examination through a skilled psychiatrist. The delirium rating scale in its original and revised version (DRS-R-98) sticks very closely to the various possibilities of psychopathological disturbances that can be found in delirium and tries to operationalize this psychopathological examination [58, 59]. Therefore, it is time-consuming and relatively complex to administer.

In 1990 the first version of the confusion assessment method (CAM) was introduced, which is based on four core criteria [2, 29]. These criteria were derived from the Diagnostic and Statistical Manual of Mental Disorders, 3rd edition, revised (DSM-III-R), a literature review as well as the discussion of an expert panel. Fig. 1 shows the CAM diagnostic algorithm in a schematic way.

Since its introduction CAM-based delirium diagnosis has become the gold standard of delirium detection. Many subsequently developed assessment tools focused on inattention and acute onset/fluctuating course as core diagnostic criteria, as defined in the DSM‑5 in 2013 [3]. Also, the upcoming ICD-11 shifts its focus towards these criteria for delirium definition [33]. The assessment of other cognitive disturbances, such as logical thinking was skipped as it was too complex and less valid. Table 1 compares core diagnostic criteria and involved psychopathological findings of delirium according to ICD-10 and ICD-11 (Table 1; [26, 27]). The comparison shows how the definition of delirium has developed from a heterogeneous psychopathological syndrome to a more specific diagnosis based primary on temporal onset, vigilance and attention, so that the needs of acute and emergency care physicians in daily routine can be better addressed.

This review, therefore, presents a summary of the available evidence on the broad variety of delirium assessments and provides suggestions on delirium detection tools that might be particularly useful in an acute care setting.

A total of 13 tools belonging to the CAM family were identified, all focusing on the 4 core symptoms: acute onset and/or fluctuating course, inattention, disorganized thinking and altered level of consciousness.

All CAM derivatives share the underlying diagnostic algorithm but differ in their structure and targeted group of persons. Most instruments are designed for patient interviews supplemented by collateral history from relatives, staff and medical records, e.g. CAM, CAM-severity scale (CAM-S), brief CAM (bCAM) and CAM for the intensive care unit (CAM-ICU) [11, 20, 30]. They often include an informal interview by which the presence of the core symptoms is evaluated. An example for a clearly structured variant is the 3‑minute diagnostic CAM (3D-CAM) with 10 questions to be answered by the patient, 10 questions to be answered by the interviewer and 2 optional questions requiring collateral history provided by relatives, friends and/or caregivers and medical records examination. Other CAM instruments are based on patient observation, e.g. nursing home CAM (NH-CAM) or caregiver interview, e.g. family CAM (FAM-CAM) [9, 51]. Furthermore, several variants address specific subgroups of patients, such as the CAM-ICU, which is also applicable to intubated patients, or the CAM for the emergency department (CAM-ED) [10, 18].

The CAM and its derivatives have many advantages but also some disadvantages. Advantages are the short examination duration, the structured evaluation algorithm and the convincing results of numerous validation studies with sensitivity and specificity values of 90–100% for delirium diagnosis compared to extensive and time-consuming clinical diagnostics; however, an important disadvantageous aspect is that for correct application of most CAM instruments intensive training is required to ensure valid and reliable test results. Some studies demonstrated poor sensitivity when the CAM was conducted by untrained or insufficiently trained raters [47, 48]. For example, the symptom disorganized thinking requires an experienced assessor for its detection. This issue has been addressed by deriving the 3D-CAM with more operationalized features [38]. In future this limitation might be overcome as the new definitions of delirium according to DSM‑5 and ICD-11 no longer include the symptom of impaired thinking.

For the emergency department and many other clinical settings there is a need for easy and quick screening tools. Examples for such assessments that could be integrated in daily clinical routine are the 4 As test (4AT) and the nursing delirium screening scale (Nu-DESC) [5, 15]. The 4AT has recently been evaluated against the CAM showing significant better sensitivity in a multicenter study [50].

As inattention is a characteristic delirium symptom many assessments that address this aspect have been evaluated in acute care and emergency settings [60]. The most popular of these is the months of the year backwards test (MOTYB) that provides a very quick option to screen for this feature. It is included in many assessments that combine testing for different aspects of delirium, such as the CAM. Several studies have evaluated its singular use in the context of delirium reporting high sensitivity and specificity [24]. Efforts have been made to further simplify the scoring method while maintaining good psychometric properties [23]. Another instrument, the bedside confusion scale (BCS), includes the MOTYB and adds one more question to assess a second delirium feature: psychomotor disturbance [52].

There are also easy to use tools focusing on the core feature acute onset. An example is the single question in delirium (SQID), which determines acute confusion by asking a relative or caregiver: “Do you feel that [patient’s name] has been more confused lately?” [48]. The SQiD can be used as a screening as well as a monitoring instrument for hospital staff.

An example for a quick tool examining the vigilance/level of consciousness of a patient, the change of which is another core feature of delirium, is the Richmond agitation sedation scale (RASS) [21]. It is widely used in intensive care units. The instrument assesses the vigilance on a scale from −5 (unarousable) to +4 (combative). There is a modified version, the mRASS, which has been optimized by adding the level of attention to each category [7].

There have been numerous attempts to use other tests originally designed for dementia diagnostics as delirium detection tools, for example the clock drawing test (CDT), but the studies revealed poor accuracy for this purpose, independent of the scoring method [1]. Other examples are the short portable mental status questionnaire (SPMSQ) or the abbreviated mental test (AMT), which are suitable to detect early dementia symptoms, but not to distinguish between dementia and delirium [8, 12].

While most of the identified tools serve solely as screening tools, some are also appropriate for monitoring the course of delirium. This group primarily comprises instruments including severity scoring and thus offering the possibility to observe change including the effect of measures taken. Examples are the CAM‑S, Nu-DESC or delirium observation screening scale (DOSS) [15, 30, 49].

The focus on inattention most delirium assessments share seems feasible for different reasons: inattention is a typical feature of delirium but not dementia, and is easy and fast to evaluate in any acute care setting [14, 27]. In dementia inattention is typically not altered unless disease progresses to severe stages. Nevertheless, in non-Alzheimer’s dementias, such as Parkinson’s disease dementia, inattention might be a cognitive domain altered in earlier stages [6].

Acute change in cognitive function is another important feature to discriminate delirium from dementia. In contrast to delirium, dementia is characterized by a chronic progressive disease course. Therefore, a second focus of many delirium assessments is the documentation of acute onset and/or fluctuating course, a central feature of the CAM family; however, a main limitation is the necessity to obtain collateral history, particularly difficult in emergency situations.

Assessments like the UB-CAM or the delirium triage screen (DTS) address this limitation by combining a short screening with a focus on inattention, followed by a second assessment that includes collateral history [20, 40]. The 4AT, that enjoys growing popularity, contains the item acute onset/fluctuating course but its presence is not mandatory for positive delirium screening [50, 57].

In order to better discriminate delirium from DSD tools have been validated to address pre-existing cognitive decline by a structured collateral history. Good psychometric properties have been shown for the informant questionnaire on cognitive decline in the elderly (IQCODE) and the Alzheimer’s disease 8 (AD8) [14, 32].

Psychomotor change has been added to the CAM algorithm to improve specificity in discriminating delirium and dementia [55]. In addition, the combination of a letter recognition attention test with the observational scale of level of arousal (OSLA) improved test performance compared to both individual tests [46].

Further frequent psychopathologic findings found in delirium are delusions and hallucinations. They are usually not integrated into commonly used assessments because they lack sensitivity and specificity for delirium diagnosis although they often have therapeutic consequences. Antipsychotic treatment should, however, only be used for very distressing delusions and hallucinations [31].

Several other diagnostic techniques have been evaluated for delirium diagnostics. Electroencephalography (EEG) can visualize alterations in brain functioning that are common in delirium, and shows potential even in differentiating delirium from dementia [54]. Recently a study described a short 1‑min single-channel EEG with new automated power analysis as a potential delirium detection tool [42]. Besides this example most EEG studies need experienced staff to obtain valid results and are, therefore, currently not implemented on a large scale. Neuroimaging, such as functional magnetic resonance imaging (fMRI) that detects network disintegration, or fluorodeoxyglucose positron emission tomography (FDG-PET) that shows focal hypometabolism in the posterior cingulate cortex, an important region for attention, can elucidate further pathophysiology but are also costly and complex to administer [19, 39]. Actimetry or other digitalized assessments are still in their infancy; however, 14 of 16 patients with CAM-positive delirium in the Canadian PredicT study were able to use a tablet-based “serious game” tapping on targets, showing digital assessments as a feasible approach [35]. A multicenter proof of assessing inattention is pursued by the DelApp. This smartphone app tests attention by presentation and recall of a sequence of symbols [56]. Other smartphone applications offer digital versions of common interview-based assessment tools facilitating documentation and practicability of the assessments [4].