Diagnosis and Treatment of Delirium in 2006

The preceding decade yielded a worldwide surge in studies of delirium with improving methodology. This progress was possible because of the robust foundation prepared by investigators such as Engel, Lipowski, Trzepacz, and Inouye. The next 10 years are primed to deliver a striking evolution in the care of patients with delirium.

The current DSM-IV-TR model portrays delirium as an acute, reversible neuropsychiatric syndrome caused by medical illness and/or drug-related activity.¹ A recent survey confirmed that many clinicians believe delirium to be transient and lacking in long-term consequences for the brain.² This belief, however, is in conflict with emerging evidence on several fronts. For one, delirium has been shown to markedly and independently affect patient outcomes such as length of stay and loss of independent living, among others.^3-8 Equally striking are studies suggesting new and persistent cognitive deficits,^4,9-14 which may be linked to known^15,16 and hypothesized^{17- 21} pathologic processes occurring within the delirious brain.

Delirium is epidemic among hospitalized patients, especially in the elderly and chronically ill, with rates reported between 12% and 80% depending on the subpopulation. Because of its pervasiveness and its associated risks for morbidity and mortality, it is imperative that delirium is diagnosed accurately, prevented, and treated.^22-24

Clinical diagnosis remains the mainstay in practice. A DSM-IV-TR diagnosis requires an acute disturbance in consciousness, new cognitive dysfunction or perceptual anomalies, and a tendency to fluctuate. Subspecifiers are:

• Delirium caused by a general medical condition.
• Substance intoxication or withdrawal delirium.
• Delirium resulting from multiple causes.¹

Inconsistency abounds in the published and clinical language as to what distinguishes a cause from a risk factor for delirium. This arbitrariness is reflected in the current DSM and in the 1999 American Psychiatric Association treatment guidelines.²⁵ The phrases "cause," "risk factor," "associated conditions," and "underlying etiology" have been and often still are used interchangeably and without scientific precision. The cerebral etiopathology of delirium has yet to be sufficiently elucidated, and thus we cannot identify a cause. The term "precipitant," however, can justifiably be used to subsume risk factors that are generally transient or acute (eg, a urinary tract infection). Similarly, "baseline vulnerability" is a term coined by Inouye²⁶ to describe the risk factors that are, by definition, chronic and innate to the patient (eg, dementia). Thus, it is proposed that there are numerous and widely varying precipitants that can activate delirium (or acute brain failure according to Lipowski) in susceptible patients (those with high baseline vulnerability).

Inouye and Charpentier26 eloquently supported this concept in their landmark 1996 study. They separated out baseline risks present at admission (eg, previous cognitive impairment) from precipitants affecting the patient after admission (eg, new-onset respiratory insufficiency). Robust patients with less baseline vulnerability ("more cerebral reserve") were more resilient in the face of new precipitants after admission. The reverse was true as well. The more baseline vulnerability a patient had, the higher the likelihood of delirium developing if his or her frail homeostasis ("less cerebral reserve") was stressed with additional precipitants. Hence, the brain-delirium relationship was demonstrated to behave much like other mechanisms of organ failure.

This would suggest that comorbid disease states and even medications may precipitate, but do not necessarily cause, delirium in vulnerable patients. The vast number and disparate nature of the identified precipitants innately argue against their having direct causal links to delirium. Exceptions may exist, of course, in which select precipitants have direct causative links, such as medications with hefty anticholinergic or pro-dopaminergic activity. The question remains how and where to fit delirium tremens and sedative withdrawal states into the larger delirium paradigm. Behavioral similarities are manifested between these clinical states and "regular" delirium, yet neurochemistry, electroencephalographic (EEG) patterns, and treatments are seemingly disparate.

The EEG is useful in the uncommon situation in which one is trying to distinguish delirium from other psychiatric states, such as catatonia, conversion disorder, or malingering. EEG findings in delirium (not delirium tremens or benzodiazepine withdrawal) consistently show diffuse disorganization and flattening, followed by reorganization at frequencies slower than normal for that person.^27,28

The reader is referred to the excellent review by Smith and colleagues²⁹ for more details on the numerous instruments that have been developed to diagnose delirium objectively in research and clinical settings. The most robust of these instruments are described below.

The Delirium Rating Scale-Revised (DRS-R-98) is a 16-item, 46-point clinician- rated scale.³⁰ It can serve to diagnose, rate the severity of, and track delirium changes over time. The best cutoff score is 15 for the severity subscale and 18 for the total score (sensitivity and specificity in the mid-90 percentiles). It has excellent internal validity and the capacity to discriminate delirium from dementia. This scale is an upgraded version of the original DRS,³¹ which lacked expanded functionality for repeated measurements or for discriminating motoric subtypes (hypoactive or hyperactive).

The Confusion Assessment Method (CAM) is designed for detection of delirium by nonpsychiatrists.³² It is an algorithm- based tool that operationalized DSMdiagnostic criteria. Sensitivity and specificity are both above 90%, and interrater reliability between trained lay interviewers and experts is high.³³ The 5-minute CAM-ICU³⁴ can be used for ventilated patients, and had a sensitivity of 73% and a specificity of 100% in one validation study.³⁵

The Memorial Delirium Assessment Scale (MDAS) of Breitbart and associates³⁶ is a 10-item, 30-point clinician scale that is best suited for rating severity. A cutoff score of 13 gives a sensitivity of 71% and a specificity of 94% for delirium. The MDAS correlates very well with the DRS and clinician's severity ratings, and the scale is well suited for serial administration. It is often used as a severity scale, with the CAM or the DRS serving to objectify the delirium diagnosis.

The Nursing Delirium Screening Scale is an intriguing instrument still in development.³⁷ This 1-minute, 5- item, 10-point nurse-scored instrument has shown a sensitivity and specificity of 86% at a 2-point cutoff, and has excellent potential for repeated or "continuous" measurement. The results of this test also correlate well with those of the MDAS and DSM-IV. This scale still needs further evaluation in older patients and for its ability to differentiate delirium from dementia.

With regard to the Mini-Mental State Examination (MMSE),³⁸ Ross and colleagues³⁹ reported the mean MMSE score to be 14.3 for delirious patients versus 29.6 for controls. The first elements of the MMSE in their relatively young cohort to show deficits were the reverse calculation, orientation, and recall items. A single MMSE is not sensitive (33%) for identifying delirium⁴⁰ and is incapable of discriminating delirium from dementia. Serial MMSEs, on the other hand, can help identify improvement or worsening of delirium, as demonstrated in a 2005 study by O'Keefe and coworkers,⁴¹ and assist in delirium screening when baseline MMSE scores are known.⁴²

A useful construct to guide delirium management in 2006 is the precipitantvulnerability- brain failure model presented earlier. After the diagnosis is made, the first task of the psychiatrist is to avoid recommending that the primary team "search for an underlying cause of delirium" and to avoid automatically recommending antipsychotics if the patient is agitated. Our first duty is to be a diligent observer and to appreciate where the patient is at clinically.

Once we are better aware of our patient's situation, we can more selectively selectively seek to address modifiable precipitants, and to decide whether and which pharmacologic agents would best serve the patient. Modifiable precipitants may, for example, be an undetected urinary tract infection, pneumonia, organ failure, sepsis, or a host of medications. Anticholinergics, benzodiazepines, corticosteroids, powerful dopamine(Drug information on dopamine) agonists, and certain opioids are best limited when feasible,^43,44 although some uncertainty remains in this arena.⁴⁵Tact and diplomacy are critical skills for the psychiatrist to exercise when addressing these precipitants with the patient's primary care physician. In the end, the goal of treating delirium is not to control agitation or hallucinations alone, but to reverse the delirium and thus mitigate associated morbidity and mortality risks.

Several comprehensive, primarily nonpharmacologic intervention protocols have been published. Regrettably, only a few of these reports are methodologically sound. The largest and bestdesigned is the Elder Life Program of Inouye and associates.⁴⁶ In this prevention program, they selected 426 nondelirious patients at risk for delirium and sought to address baseline cognitive impairment, sleep, mobility, vision, hearing, and dehydration. They reported a 5% decrease in delirium incidence compared with a usual-care control group. One year after discharge, the intervention group also demonstrated a 15% reduction in costs and length of stay at nursing homes compared with the controls.⁴⁷

Naughton and colleagues⁴⁸ reported reducing delirium incidence from 41% to 19%, and reducing length of stay for delirious patients in a specialty geriatrics unit by means of their intervention program. This program sought to maximize provider coordination, knowledge, and awareness, and used both medication-based and non- medication-based interventions. The raters in this study were not blind to the treatment group.

Using the less specific Organic Brain Syndrome Scale, an early prevention study described a reduction in delirium incidence, duration, severity, and hospital length of stay for postoperative patients.⁴⁹ This protocol consisted chiefly of oxygen therapy and prevention of perioperative hypotension. On the other hand, one group found their comprehensive nonpharmacologic intervention program to be ineffective in reducing delirium incidence.⁵⁰

Overall, environmental and nursing measures have been suggested to have an important role in delirium management, particularly if smartly targeted at modifiable precipitants; these measures are inclusive of delirium recognition and education-promoting components.