Delirium must be diagnosed accurately, prevented, and treated because of its pervasiveness and the associated risks of morbidity and mortality.
The preceding decade yielded aworldwide surge in studies ofdelirium with improving methodology.This progress was possiblebecause of the robust foundationprepared by investigators such as Engel,Lipowski, Trzepacz, and Inouye. Thenext 10 years are primed to deliver astriking evolution in the care of patientswith delirium.
The current DSM-IV-TR modelportrays delirium as an acute, reversibleneuropsychiatric syndrome caused bymedical illness and/or drug-relatedactivity.1 A recent survey confirmedthat many clinicians believe deliriumto be transient and lacking in long-termconsequences for the brain.2 This belief,however, is in conflict with emergingevidence on several fronts. For one, deliriumhas been shown to markedly andindependently affect patient outcomessuch as length of stay and loss of independentliving, among others.3-8 Equallystriking are studies suggesting new andpersistent cognitive deficits,4,9-14 whichmay be linked to known15,16 and hypothesized17-21 pathologic processes occurringwithin the delirious brain.
Delirium is epidemic among hospitalizedpatients, especially in the elderlyand chronically ill, with rates reportedbetween 12% and 80% depending onthe subpopulation. Because of its pervasivenessand its associated risks formorbidity and mortality, it is imperativethat delirium is diagnosed accurately,prevented, and treated.22-24
Clinical diagnosis remains the mainstayin practice. A DSM-IV-TR diagnosisrequires an acute disturbance inconsciousness, new cognitive dysfunctionor perceptual anomalies, and atendency to fluctuate. Subspecifiers are:
Inconsistency abounds in the publishedand clinical language as to whatdistinguishes a cause from a risk factorfor delirium. This arbitrariness isreflected in the current DSM and in the1999 American Psychiatric Associationtreatment guidelines.25 The phrases "cause," "risk factor," "associated conditions,"and "underlying etiology" havebeen and often still are used interchangeablyand without scientific precision.The cerebral etiopathology ofdelirium has yet to be sufficiently elucidated,and thus we cannot identify acause. The term "precipitant," however,can justifiably be used to subsume risk factors that are generally transient oracute (eg, a urinary tract infection).Similarly, "baseline vulnerability" is aterm coined by Inouye26 to describe therisk factors that are, by definition,chronic and innate to the patient (eg,dementia). Thus, it is proposed that thereare numerous and widely varyingprecipitants that can activate delirium(or acute brain failure according toLipowski) in susceptible patients (thosewith high baseline vulnerability).
Inouye and Charpentier26 eloquentlysupported this concept in their landmark1996 study. They separated out baselinerisks present at admission (eg,previous cognitive impairment) fromprecipitants affecting the patient afteradmission (eg, new-onset respiratoryinsufficiency). Robust patients with lessbaseline vulnerability ("more cerebralreserve") were more resilient in the faceof new precipitants after admission.The reverse was true as well. The morebaseline vulnerability a patient had, thehigher the likelihood of delirium developingif his or her frail homeostasis("less cerebral reserve") was stressedwith additional precipitants. Hence, thebrain-delirium relationship was demonstratedto behave much like other mechanismsof organ failure.
This would suggest that comorbiddisease states and even medicationsmay precipitate, but do not necessarilycause, delirium in vulnerable patients.The vast number and disparate natureof the identified precipitants innatelyargue against their having direct causallinks to delirium. Exceptions may exist,of course, in which select precipitantshave direct causative links, such asmedications with hefty anticholinergicor pro-dopaminergic activity. The question remains how and where to fit deliriumtremens and sedative withdrawalstates into the larger delirium paradigm.Behavioral similarities are manifestedbetween these clinical states and"regular" delirium, yet neurochemistry,electroencephalographic (EEG) patterns,and treatments are seeminglydisparate.
The EEG is useful in the uncommonsituation in which one is trying to distinguishdelirium from other psychiatricstates, such as catatonia, conversiondisorder, or malingering. EEG findingsin delirium (not delirium tremens orbenzodiazepine withdrawal) consistentlyshow diffuse disorganization andflattening, followed by reorganizationat frequencies slower than normal forthat person.27,28
The reader is referred to the excellentreview by Smith and colleagues29 formore details on the numerous instrumentsthat have been developedto diagnose delirium objectively inresearch and clinical settings. Themost robust of these instruments aredescribed 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 trackdelirium changes over time. The bestcutoff score is 15 for the severity subscaleand 18 for the total score (sensitivityand specificity in the mid-90percentiles). It has excellent internalvalidity and the capacity to discriminatedelirium from dementia. This scaleis an upgraded version of the originalDRS,31 which lacked expanded functionalityfor repeated measurements orfor discriminating motoric subtypes(hypoactive or hyperactive).
The Confusion Assessment Method(CAM) is designed for detection of deliriumby nonpsychiatrists.32 It is an algorithm-based tool that operationalizedDSMdiagnostic criteria. Sensitivity and specificity are both above 90%, andinterrater reliability between trained layinterviewers and experts is high.33 The5-minute CAM-ICU34 can be used forventilated patients, and had a sensitivityof 73% and a specificity of 100%in one validation study.35
The Memorial Delirium AssessmentScale (MDAS) of Breitbart and associates36is a 10-item, 30-point clinicianscale that is best suited for rating severity.A cutoff score of 13 gives a sensitivityof 71% and a specificity of 94%for delirium. The MDAS correlates verywell with the DRS and clinician's severityratings, and the scale is well suitedfor serial administration. It is often usedas a severity scale, with the CAM orthe DRS serving to objectify the deliriumdiagnosis.
The Nursing Delirium ScreeningScale is an intriguing instrument stillin development.37 This 1-minute, 5-item, 10-point nurse-scored instrumenthas shown a sensitivity and specificityof 86% at a 2-point cutoff, and has excellentpotential for repeated or "continuous"measurement. The results of thistest also correlate well with those of theMDAS and DSM-IV. This scale stillneeds further evaluation in older patientsand for its ability to differentiate deliriumfrom dementia.
With regard to the Mini-Mental StateExamination (MMSE),38 Ross andcolleagues39 reported the mean MMSEscore to be 14.3 for delirious patientsversus 29.6 for controls. The first elementsof the MMSE in their relativelyyoung cohort to show deficits were thereverse calculation, orientation, andrecall items. A single MMSE is notsensitive (33%) for identifying delirium40and is incapable of discriminatingdelirium from dementia. SerialMMSEs, on the other hand, can helpidentify improvement or worsening ofdelirium, as demonstrated in a 2005study by O'Keefe and coworkers,41 andassist in delirium screening when baselineMMSE scores are known.42
A useful construct to guide deliriummanagement in 2006 is the precipitantvulnerability-brain failure model presentedearlier. After the diagnosis ismade, the first task of the psychiatristis to avoid recommending that theprimary team "search for an underlyingcause of delirium" and to avoidautomatically recommending antipsychoticsif the patient is agitated. Ourfirst duty is to be a diligent observerand to appreciate where the patient isat clinically.
Once we are better aware of ourpatient's situation, we can more selectively selectivelyseek to address modifiable precipitants,and to decide whether and whichpharmacologic agents would best servethe patient. Modifiable precipitantsmay, for example, be an undetectedurinary tract infection, pneumonia,organ failure, sepsis, or a host of medications.Anticholinergics, benzodiazepines, corticosteroids, powerfuldopamine agonists, and certain opioidsare best limited when feasible,43,44although some uncertainty remains inthis arena.45Tact and diplomacy are criticalskills for the psychiatrist to exercisewhen addressing these precipitantswith the patient's primary care physician.In the end, the goal of treatingdelirium is not to control agitation orhallucinations alone, but to reverse thedelirium and thus mitigate associatedmorbidity and mortality risks.
Several comprehensive, primarily nonpharmacologicintervention protocolshave been published. Regrettably, onlya few of these reports are methodologicallysound. The largest and bestdesignedis the Elder Life Program ofInouye and associates.46 In this preventionprogram, they selected 426nondelirious patients at risk for deliriumand sought to address baselinecognitive impairment, sleep, mobility,vision, hearing, and dehydration. Theyreported a 5% decrease in delirium incidencecompared with a usual-carecontrol group. One year after discharge,the intervention group also demonstrateda 15% reduction in costs andlength of stay at nursing homes comparedwith the controls.47
Naughton and colleagues48 reportedreducing delirium incidence from 41%to 19%, and reducing length of stay fordelirious patients in a specialty geriatricsunit by means of their interventionprogram. This program sought tomaximize provider coordination,knowledge, and awareness, and usedboth medication-based and non- medication-based interventions. The raters in this study were not blind tothe treatment group.
Using the less specific Organic BrainSyndrome Scale, an early preventionstudy described a reduction in deliriumincidence, duration, severity, and hospitallength of stay for postoperativepatients.49 This protocol consisted chiefly of oxygen therapy and preventionof perioperative hypotension. Onthe other hand, one group found theircomprehensive nonpharmacologic interventionprogram to be ineffective inreducing delirium incidence.50
Overall, environmental and nursingmeasures have been suggested to havean important role in delirium management,particularly if smartly targeted atmodifiable precipitants; these measuresare inclusive of delirium recognition andeducation-promoting components.
Pharmacologic options for delirium areemerging as powerful options beyondenvironmental and nursing measures.The clinician's first task, though, is toseek clarity of what is to be treated.
The 2 most accepted neurotransmitterchanges in delirium (not deliriumtremens) are a reduction in acetylcholine(ACH) activity and an excess of dopamine(DA) activity.51ACH has long beenknown to be decreased in patients withdelirium.52 Excess DA agonism can leadto delirium, as seen with drugs like Ldopaor cocaine. Agonists of DA havealso been shown to cause EEG slowingdespite motoric hyperactivity.53Moreover, there is interplay betweenDA and ACH, as shown by D2 antagonistsenhancing ACH release54 andclinically by the usefulness of DA antagonistsin reversing delirium precipitatedby anticholinergics.52
DA antagonists in the form of antipsychoticshave been used for decadesas the treatment of choice for delirium.Haloperidol continues to be used widelybecause of its flexible dosage forms,relative safety, and familiarity to physicians.Intravenous haloperidol offers onset of action (calming effects) within5 to 20 minutes, while the effects of theoral form peak in about 4 hours.55Resolution of delirium can take 1 to 5days, with response rates in the 70% to80% range. Typical dosages are 1 to 10mg/d, often divided bid, or more.
Haloperidol was used as a prophylaxisagainst delirium in 2 publishedtrials, although both have methodologicshortcomings. The first study reportedthat delirium developed in 10% of agroup receiving intravenous haloperidol(5 mg/d) over 4 postoperative days,compared with 32% of the controlgroup.56 The study researchers did notuse an objective rating scale, nor didthey report who rated the patients. Thesecond and better-designed study didnot find a difference in delirium incidencewith the use of oral haloperidolprophylaxis (0.5 mg tid) versusplacebo,57 but did find reduced severityand duration of delirium, as well as ashorter hospital stay, among those inthe treatment group. Data confounderswere the low dosage and comprehensivegeriatric consultations for bothstudy groups, which probably dilutedtreatment effects. Of interest, severalnew reports have noted that haloperidolhas favorable immune-modulatingeffects,58,59 which may drive the lowermortality of ventilated patients treatedwith haloperidol.60
Duration of treatment is practitionerdependent,with no data-driven guidelines.In my practice, I recommend a50% reduced dose for 1 week aftersymptoms improve for those patientswith prolonged and slowly recoveringcourses of delirium, or for those estimatedto have low cerebral reserve.Patients who fully recover quickly donot appear to require continued antipsychotics,assuming that the precipitantswere able to be mitigated.
At least 20 English-language reportsare available regarding atypical antipsychoticsfor delirium (Table). Only 1 ofthose 20 reports, however, was arandomized, double-blind trial (risperidone),and it was powered as a pilotstudy only. Specialists in psychosomaticmedicine have been successfullyusing these agents (particularly risperidone,olanzapine, and quetiapine) inclinical practice for several years despitescanty literature support and other pressures.The advantages are the loweredrisk of extrapyramidal symptoms orECG abnormalities,61 mood-modulatingeffects, and possibly, enhanced efficacyin select patients.
The concern about the in vitro antimuscarinicbinding affinity of olanzapinemay be incompletely understood.There is evidence to suggest that thenet central cholinergic activity may beagonistic via several serotonin-ACHreceptor interactions.62 This would correlate with this drug's tolerability andsuccess in treating delirium.
Amisulpride, which is not availablein the United States, has also beenreported to be as effective as quetiapinein an open trial, with an average of6 days to stabilization.63 There are 2reports of ziprasidone used in singlepatients, the second of which noted adramatic 5-minute reversal of deliriumin a patient in an ICU who took 20 mgintravenously.64,65 Intravenous ondansetronhas also been reported to a havea similar dramatic effect on deliriumreversal, but this finding awaits carefulreplication.66 Overall, there remains adearth of sound clinical trials to examinewhether there is a tiered efficacy fordelirium treatment.
Additional study is also warrantedin the treatment of hypoactive deliriumwith antipsychotic medications. Oneopen study found that low-dosehaloperidol and chlorpromazinequickly reduced DRS scores equally inboth hyperactive and hypoactivepatients with AIDS.67 Treatment maybe crucial, since this subgroup is theleast likely to have their deliriumcorrectly diagnosed, and has a worseprognosis than the hyperactive or mixedsubtypes.68
Pro-cholinergic agents. The serumanticholinergic assay (SAA) of Tune andcolleagues69 has been used to demonstratethat numerous commonly usedmedications have anticholinergic effects.SAA results also correlate with MMSEscores in community-dwelling elderlypatients,70 and delirious patients havebeen shown in several studies to havehigher SAA levels than controls.71-73 Aninteresting new and unreplicated discoveryis of unidentified endogenous anticholinergicsubstances in hospitalizedelderly persons.74 Furthermore, thereappears to be a substantial interplaybetween ACH systems in the brain andanesthetic agents.20
Armed with this knowledge, therehas been a resurgent interest in usingpro-cholingeric agents in delirium, mostnotably ACH/butyrylcholinesteraseinhibitors. Several case reports andsmall case series have noted improvementsin beclouded dementia,75 opioidprecipitateddelirium,76,77 postsurgicaldelirium,78,79 and antipsychotic-resistantdelirium.80,81 In a progressive andwell-designed study, 5 mg of donepezildaily was no more effective than placeboin preventing delirium in a relativelyhealthy, low-risk population.79 Theseagents may, nonetheless, play a prophylacticrole in reducing the risk of deliriumin vulnerable populations,82,83although we await a sound study to testthis proposition.
Other approaches. Additional hypothesizedmechanisms are emerging andmay become targets of future biologic therapies for delirium. One group useddexmedetomidine in a postoperativesedation protocol.84 They reported astunning 5% 3-day incidence of deliriumin the group receiving the α2&
151;agonist versus a 51% incidence forthose who received standard sedationprotocols. In addition, the CNS andvagus nerve–based "cholinergic antiinflammatorypathway" has become atopic of interest in the prevention ofinflammation-related cell injury, andcould have relevance for deliriumrelatedmorbidities.85
Goals for clinical practice and researchin the coming decade are to increaseawareness of the legitimate morbiditydelirium wields; to produce soundresearch in questioning etiopathogenesisand neurotoxicity; and to establishdata-driven prevention and treatmentstrategies. It will be an exciting daywhen we can routinely prevent deliriumand offer better protection for ourpatients against spiraling functionaland cognitive decline.