Review: Insufficient evidence exists about which drugs are associated with delirium; benzodiazepines may increase risk

Question

Which drugs are associated with increased risk for delirium?

Review scope

Included studies evaluated the association between drugs and delirium in hospital patients or long-term care residents. Outcome was delirium (International Classification of Diseases volume 10 or Diagnostic and Statistical Manual for Mental Disorders [DSM] volumes III, III-R, or IV criteria, or a diagnostic tool validated against those criteria).

Review methods

MEDLINE, EMBASE/Excerpta Medica, PsycInfo, and Allied and Complementary Medicine (all to Oct 2009); and reference lists were searched for randomized controlled trials (RCTs), prospective cohort studies, and case–control studies. 14 studies (n = 4652, mean age 55 to 89 y) met the selection criteria, including 3 RCTs, 10 prospective cohort studies, and 1 nested case–control study. 7 studies evaluated opioids, 7 benzodiazepines, 4 neuroleptics, 2 antihistamine H₁–antagonists, and 1 each evaluated histamine H₂–antagonists, dihydropyridines, antimuscarinincs, tricyclic antidepressants, antiparkinson medications, digoxin, steroids, and nonsteroidal antiinflammatory drugs.

Main results

Meta-analyses were not done because studies were heterogeneous in populations and methods. 1 RCT found that haloperidol and placebo did not differ for delirium (Table). Using multivariate analysis, 1 cohort study found that neuroleptics and opioids were each associated with increased risk for delirium; another cohort study did not find an association between morphine or fentanyl and delirium (Table). 1 matched case–control study found that benzodiazepines were associated with increased risk for delirium, but opioids, including morphine and fentanyl, were not (Table).

Risk for delirium associated with drugs and drug classes^*

Drug class	Drug	Study type (n)†	RR/OR (95% CI)‡
Neuroleptic	Haloperidol	High-quality, placebo- (430) controlled RCT	RR 0.9 (0.6 to 1.3)
	All types	Prospective cohort (325)	OR 4.5 (1.8 to 10.5)§
Opioid	All types	Prospective cohort (325)	OR 2.5 (1.2 to 5.2)§
		Nested case–control (1341)	OR 1.4 (0.5 to 4.3)
	Morphine	Prospective cohort (198)	OR 1.1 (0.9 to 1.2)
		Nested case–control (1341)	OR 1.2 (0.6 to 2.4)
	Fentanyl	Prospective cohort (198)	OR 1.2 (1.0 to 1.5)
		Nested case–control (1341)	OR 1.5 (0.6 to 4.2)
Benzodiazepine	All types	Nested case–control (1341)	OR 3.0 (1.3 to 6.8)§

^*OR = odds ratio; RCT = randomized controlled trial; RR = risk ratio. High-quality studies, moderate-quality studies evaluating drug classes, and single-drug comparisons evaluated in > 1 moderate-quality study are reported here.

^†Cohort and case–control studies used multivariate or matched analyses and were of moderate quality.

^‡RRs and ORs > 1 indicate increased risk for delirium associated with drug.

^§Statistically significant.

Conclusions

Benzodiazepines may be associated with increased risk for delirium. Insufficient evidence exists to evaluate other drugs.

Appendix

Commentary

Number, dose, and specific classes of medications are frequently implicated as precipitating factors for delirium, and Clegg and colleagues evaluated the risks associated with specific medications. The study has several methodological limitations. First, the search and data abstraction methods are not clear. Second, 8 trials defined delirium using the Confusion Assessment Method (CAM) or CAM-Intensive Care Unit screening instruments, which may not be as accurate as DSM criteria (1), and the authors did not provide analyses that stratified medication risk by the criterion standard used. Third, multivariate analyses were evaluated for inclusion of 3 important delirium confounders: age, cognitive impairment, and illness severity. However, health care providers often do not identify cognitive impairment, so the setting and methods of identifying predelirium dementia are not inconsequential (2). Finally, recent RCTs that were not included in the review suggest that such medications as melatonin or olanzapine may prevent incident delirium, whereas other trials provide contradictory data about the inverse relation between narcotic dose and risk for delirium (3–5).

The results of the review by Clegg and colleagues provide clinicians with an initial list of medications to avoid in patients at risk for delirium. Meperidine should be avoided because other narcotics provide analgesia with less risk for delirium. Some patient subgroups seem to have an increase in delirium with lower doses of narcotic analgesia, so the degree of pain should be assessed concurrently with the medication dose. Benzodiazepines with longer half-lives seem to precipitate delirium more frequently. However, for most medications, the risk for precipitating iatrogenic delirium is inconclusive.

Christopher R. Carpenter, MD, MSc

Washington University

St. Louis, Missouri, USA