Delirium is a vexing clinical problem in hospitalized older adults. Despite the established association between delirium and death, morbidity, institutionalization, and patient distress (Siddiqi et al., Reference Siddiqi, House and Holmes2006), few high-quality studies examining the management of delirium in older adults exist. Delirium management strategies can be broadly divided into those addressing prevention or treatment, as well as pharmacological and non-pharmacological modalities. Additional considerations include peri-operative and post-operative delirium management and management of delirium in the intensive care unit (ICU).
Among hospitalized older adults, multicomponent non-pharmacological interventions have the best evidence for primary prevention of delirium. Such strategies, ideally delivered by an interdisciplinary team, include early mobilization, pain management, infection prevention, use of vision and hearing aids, avoidance of sleep disruption, adequate hydration and nutrition, re-orientation, cognitive stimulation, and review of psychoactive medications. A meta-analysis of 14 interventional studies found that implementation of the Hospital Elder Life Program (Inouye et al., Reference Inouye1999), a multicomponent interdisciplinary intervention, reduced the risk of incident delirium by 53% (OR 0.47 [95% CI, 0.38–0.58]) among hospitalized, non-ICU patients age 65 years and older (Hshieh et al., Reference Hshieh2015). On the other hand, there is no high-quality evidence supporting the use of multi-component interventions to treat delirium once it has developed.
Despite widespread use of pharmacological treatments, particularly antipsychotics, to manage delirium in hospitalized patients, most studies show no benefit in the use of medications to either prevent or treat delirium in hospitalized older adults (outside of specific cases such as alcohol-withdrawal delirium) (Oh et al., Reference Oh, Fong, Hshieh and Inouye2017). Medications studied in randomized, placebo-controlled trials for either prevention or treatment of delirium in older adults include haloperidol, atypical antipsychotics (aripiprazole, risperidone, olanzapine, quetiapine), melatonin, ramelteon, gabapentin, pregabalin, ondansetron, donepezil, methylprednisolone (at anesthetic induction), and dexmedetomidine (in the ICU and peri-operatively) (Oh et al., Reference Oh, Fong, Hshieh and Inouye2017). Only the centrally acting alpha-2 agonist dexmedetomidine has shown benefit in reducing the incidence and duration of ICU delirium (Flükiger et al., Reference Flükiger2018).
The lack of evidence-based pharmacological treatment options presents a conundrum for clinicians managing patients with delirium, particularly those with hyperactive delirium who may have psychosis, severe agitation, and are distressed. In their systematic review published in this issue of International Psychogeriatrics, Gupta et al. (Reference Gupta2022) note that anticonvulsants are used clinically to manage delirium, with observational studies suggesting some benefit in the use of valproic acid to treat delirium. However, as the authors note, these studies were conducted in adults across the lifespan and the findings may not generalize to older adults, the population at highest risk of delirium. In order to critically appraise the evidence for this class of medication in managing delirium in older adults, they performed a systematic review with the goal of identifying randomized controlled trials (RCTs) examining the use of anticonvulsants to prevent or treat delirium in this population.
Gupta et al. (Reference Gupta2022) identified findings from only four RCTs (two primary analyses and two post-hoc analyses) in their review, all involving gabapentinoids (pregabalin or gabapentin), and all in patients undergoing orthopedic or spinal surgeries. The putative mechanism of gabapentinoids in delirium prevention in this population is primarily via opioid-sparing (Fabritius et al., Reference Fabritius2017). This is in contrast to the hypothetical mechanism of valproic acid in preventing and treating delirium, which is not well elucidated, but is postulated to be through neuromodulatory effects resulting in reduced inflammation, oxidative stress, and transcriptional changes that are implicated in the pathophysiology of delirium (Sher et al., Reference Sher, Miller Cramer, Ament, Lolak and Maldonado2015).
Of the four studies that Gupta et al. (Reference Gupta2022) identified, only one has positive findings: a pilot double-blind, placebo-controlled RCT with the primary aim of examining the use of gabapentin versus placebo in addition to opioid medication for post-operative pain management in 21 patients aged 45 years and older undergoing spinal surgery. In this study, the authors found that zero out of nine participants receiving gabapentin experienced delirium, while five out of 12 patients in the placebo group experienced delirium, a statistically significant difference (Leung et al., Reference Leung2006). However, in a larger, follow-up double-blind placebo-controlled RCT by the same research group examining adults aged 65 years and older undergoing spinal surgery or hip or knee arthroplasty, the same intervention, while found to reduce opioid use post-operatively, did not reduce the rate, severity or duration of delirium compared to placebo (Leung et al., Reference Leung2017).
The other two studies were post-hoc analyses of double-blind, placebo-controlled RCTs investigating the use of gabapentin and pregabalin, respectively, in pain control following orthopedic surgery (Dighe et al., Reference Dighe, Clarke, McCartney and Wong2014; Farlinger et al., Reference Farlinger, Clarke and Wong2018). Neither study found a difference in the incidence of delirium between the active and placebo groups. Of note, however, the incidence of delirium was very low in one of the studies, at 0.6% (Farlinger et al., Reference Farlinger, Clarke and Wong2018), and participants in both studies were relatively young, with mean ages of 62.7 and 60.1 years, respectively (Dighe et al., Reference Dighe, Clarke, McCartney and Wong2014; Farlinger et al., Reference Farlinger, Clarke and Wong2018). These two studies were from the same group of authors, meaning that all the literature in this area comes from the same two research groups.
As Gupta et al. (Reference Gupta2022) point out in their conclusion, there is no high-quality evidence supporting the use of anticonvulsants in preventing or treating delirium in hospitalized older adults. Their review highlights some challenges in the delirium literature as a whole. As discussed above, the best evidence for delirium management comes from multicomponent prevention studies, whereas treatment studies have generally yielded negative results. There are a number of factors that could contribute to these null findings.
First, the diagnosis of delirium is not always straightforward, particularly in patients with dementia, and diagnostic heterogeneity in clinical trials might dilute any potential treatment effect. A number of instruments have been developed to screen for and diagnose delirium. The Confusion Assessment Method (CAM), based on the DSM criteria, is one of the most commonly used instruments in clinical practice and research (Inouye et al., Reference Inouye, van Dyck, Alessi, Balkin, Siegal and Horwitz1990; Shi et al., Reference Shi, Warren, Saposnik and Macdermid2013). The CAM involves both direct observation and a brief cognitive assessment. It has been shown to have good psychometric evidence for diagnosing delirium in a variety of medical settings, with a pooled sensitivity of 82% and specificity of 99% when compared to the gold-standard DSM criteria in meta-analysis (Shi et al., Reference Shi, Warren, Saposnik and Macdermid2013). Although the CAM has not been studied extensively in patients with dementia, existing evidence from small studies suggests that it has adequate diagnostic accuracy in this population (Morandi et al., Reference Morandi2012).
Inouye’s chart abstraction tool (used in two of the studies identified by Gupta et al., Reference Gupta2022) has been shown to have a sensitivity of 74% and specificity of 83% using the CAM as the gold standard (Inouye et al., Reference Inouye, Leo-Summers, Zhang, Bogardus, Leslie and Agostini2005). However, dementia has been shown to be associated with false positive delirium ratings using this chart abstraction tool, with a dementia diagnosis present in 20% of the false positive ratings (vs. 9% of the true negative ratings) (Inouye et al., Reference Inouye, Leo-Summers, Zhang, Bogardus, Leslie and Agostini2005). Thus, dementia, which is a significant risk factor for delirium, has the potential to confound the outcomes of clinical trials in which the diagnosis of delirium was made via retrospective chart review.
The most straightforward potential reason for negative pharmacological studies is that existing treatments are simply not effective. Delirium is often multi-factorial, particularly in older adults with multimorbidity and polypharmacy, and a single medication is likely often not enough to address all of the contributing variables. For example, gabapentinoids may reduce opioid use, but do not address other potential peri-operative contributors to delirium, such as sleep disruption, anesthetic agents, reduced mobility, and dehydration. Alternatively, medication treatment may improve some delirium symptoms, but not enough to change patients’ overall outcome based on standardized delirium rating scales. For example, antipsychotics can potentially reduce agitation and psychosis and help to consolidate sleep, but do not necessarily improve cognitive performance and awareness (or could make these features worse). This variability in delirium symptom response might be particularly pertinent for persons with dementia who have cognitive and behavioral symptoms at baseline, making it more difficult to measure treatment effect.
The most studied medications for delirium treatment are antipsychotics, primarily haloperidol, and second-generation antipsychotics. However, despite widespread clinical use, existing evidence suggests that antipsychotics are no better than placebo in reducing delirium severity or mortality (Burry et al., Reference Burry2018). In their Cochrane review examining the use of antipsychotics to treat delirium in hospitalized non-ICU patients, Burry et al. not only identified the overall null findings but also noted the majority of the RCTs in this area to be of low methodological quality (Burry et al., Reference Burry2018), meaning that one cannot be confident in the findings. The authors identified methodological limitations such as inadequate sample sizes, attrition bias, variable outcome measures (e.g., delirium severity versus delirium resolution) and measurement instruments, and inadequate reporting of use of additional non-pharmacological strategies, sedating medications (e.g., benzodiazepines), and physical restraints. They also noted that studies did not generally report on adverse events or length of hospital stay or hospital disposition, potentially clinically important outcome variables.
Fortunately, several initiatives are underway to standardize delirium research. The Network for Investigation of Delirium: Unifying Scientists (NIDUS) (https://deliriumnetwork.org) is working to develop a core outcome set of patient-centered outcomes and evidence-based measurement instruments for use in delirium research (Rose et al., Reference Rose2017). The International Federation of Delirium Societies (iDelirium; http://www.idelirium.org/) has proposed a diagnostic framework to improve the diagnosis of delirium superimposed on dementia (Morandi et al., Reference Morandi2017). In addition to standardization of outcomes and measurement instruments in delirium clinical trials, adequately powered prospective placebo-controlled RCTs, particularly in hospitalized non-ICU patients (the most common, but least studied patient group), are needed to determine the efficacy of delirium treatment options. Further, as suggested by Burry et al., clinical trials should evaluate the impact of medication administered in the context of standardized non-pharmacological delirium prevention and treatment strategies (Burry et al., Reference Burry2018). Ideally, clinical trials would include pre-specified subgroup analyses examining the efficacy of interventions in patients with dementia, as they are both at high risk for delirium and potentially less likely to respond to treatment (and to have more adverse effects) than persons with less neurological vulnerability. Hopefully, with high-quality, interdisciplinary and collaborative research, patients with delirium will have better treatment options in the coming years.
You have
Access
Delirium is a vexing clinical problem in hospitalized older adults. Despite the established association between delirium and death, morbidity, institutionalization, and patient distress (Siddiqi et al., Reference Siddiqi, House and Holmes2006), few high-quality studies examining the management of delirium in older adults exist. Delirium management strategies can be broadly divided into those addressing prevention or treatment, as well as pharmacological and non-pharmacological modalities. Additional considerations include peri-operative and post-operative delirium management and management of delirium in the intensive care unit (ICU).
Among hospitalized older adults, multicomponent non-pharmacological interventions have the best evidence for primary prevention of delirium. Such strategies, ideally delivered by an interdisciplinary team, include early mobilization, pain management, infection prevention, use of vision and hearing aids, avoidance of sleep disruption, adequate hydration and nutrition, re-orientation, cognitive stimulation, and review of psychoactive medications. A meta-analysis of 14 interventional studies found that implementation of the Hospital Elder Life Program (Inouye et al., Reference Inouye1999), a multicomponent interdisciplinary intervention, reduced the risk of incident delirium by 53% (OR 0.47 [95% CI, 0.38–0.58]) among hospitalized, non-ICU patients age 65 years and older (Hshieh et al., Reference Hshieh2015). On the other hand, there is no high-quality evidence supporting the use of multi-component interventions to treat delirium once it has developed.
Despite widespread use of pharmacological treatments, particularly antipsychotics, to manage delirium in hospitalized patients, most studies show no benefit in the use of medications to either prevent or treat delirium in hospitalized older adults (outside of specific cases such as alcohol-withdrawal delirium) (Oh et al., Reference Oh, Fong, Hshieh and Inouye2017). Medications studied in randomized, placebo-controlled trials for either prevention or treatment of delirium in older adults include haloperidol, atypical antipsychotics (aripiprazole, risperidone, olanzapine, quetiapine), melatonin, ramelteon, gabapentin, pregabalin, ondansetron, donepezil, methylprednisolone (at anesthetic induction), and dexmedetomidine (in the ICU and peri-operatively) (Oh et al., Reference Oh, Fong, Hshieh and Inouye2017). Only the centrally acting alpha-2 agonist dexmedetomidine has shown benefit in reducing the incidence and duration of ICU delirium (Flükiger et al., Reference Flükiger2018).
The lack of evidence-based pharmacological treatment options presents a conundrum for clinicians managing patients with delirium, particularly those with hyperactive delirium who may have psychosis, severe agitation, and are distressed. In their systematic review published in this issue of International Psychogeriatrics, Gupta et al. (Reference Gupta2022) note that anticonvulsants are used clinically to manage delirium, with observational studies suggesting some benefit in the use of valproic acid to treat delirium. However, as the authors note, these studies were conducted in adults across the lifespan and the findings may not generalize to older adults, the population at highest risk of delirium. In order to critically appraise the evidence for this class of medication in managing delirium in older adults, they performed a systematic review with the goal of identifying randomized controlled trials (RCTs) examining the use of anticonvulsants to prevent or treat delirium in this population.
Gupta et al. (Reference Gupta2022) identified findings from only four RCTs (two primary analyses and two post-hoc analyses) in their review, all involving gabapentinoids (pregabalin or gabapentin), and all in patients undergoing orthopedic or spinal surgeries. The putative mechanism of gabapentinoids in delirium prevention in this population is primarily via opioid-sparing (Fabritius et al., Reference Fabritius2017). This is in contrast to the hypothetical mechanism of valproic acid in preventing and treating delirium, which is not well elucidated, but is postulated to be through neuromodulatory effects resulting in reduced inflammation, oxidative stress, and transcriptional changes that are implicated in the pathophysiology of delirium (Sher et al., Reference Sher, Miller Cramer, Ament, Lolak and Maldonado2015).
Of the four studies that Gupta et al. (Reference Gupta2022) identified, only one has positive findings: a pilot double-blind, placebo-controlled RCT with the primary aim of examining the use of gabapentin versus placebo in addition to opioid medication for post-operative pain management in 21 patients aged 45 years and older undergoing spinal surgery. In this study, the authors found that zero out of nine participants receiving gabapentin experienced delirium, while five out of 12 patients in the placebo group experienced delirium, a statistically significant difference (Leung et al., Reference Leung2006). However, in a larger, follow-up double-blind placebo-controlled RCT by the same research group examining adults aged 65 years and older undergoing spinal surgery or hip or knee arthroplasty, the same intervention, while found to reduce opioid use post-operatively, did not reduce the rate, severity or duration of delirium compared to placebo (Leung et al., Reference Leung2017).
The other two studies were post-hoc analyses of double-blind, placebo-controlled RCTs investigating the use of gabapentin and pregabalin, respectively, in pain control following orthopedic surgery (Dighe et al., Reference Dighe, Clarke, McCartney and Wong2014; Farlinger et al., Reference Farlinger, Clarke and Wong2018). Neither study found a difference in the incidence of delirium between the active and placebo groups. Of note, however, the incidence of delirium was very low in one of the studies, at 0.6% (Farlinger et al., Reference Farlinger, Clarke and Wong2018), and participants in both studies were relatively young, with mean ages of 62.7 and 60.1 years, respectively (Dighe et al., Reference Dighe, Clarke, McCartney and Wong2014; Farlinger et al., Reference Farlinger, Clarke and Wong2018). These two studies were from the same group of authors, meaning that all the literature in this area comes from the same two research groups.
As Gupta et al. (Reference Gupta2022) point out in their conclusion, there is no high-quality evidence supporting the use of anticonvulsants in preventing or treating delirium in hospitalized older adults. Their review highlights some challenges in the delirium literature as a whole. As discussed above, the best evidence for delirium management comes from multicomponent prevention studies, whereas treatment studies have generally yielded negative results. There are a number of factors that could contribute to these null findings.
First, the diagnosis of delirium is not always straightforward, particularly in patients with dementia, and diagnostic heterogeneity in clinical trials might dilute any potential treatment effect. A number of instruments have been developed to screen for and diagnose delirium. The Confusion Assessment Method (CAM), based on the DSM criteria, is one of the most commonly used instruments in clinical practice and research (Inouye et al., Reference Inouye, van Dyck, Alessi, Balkin, Siegal and Horwitz1990; Shi et al., Reference Shi, Warren, Saposnik and Macdermid2013). The CAM involves both direct observation and a brief cognitive assessment. It has been shown to have good psychometric evidence for diagnosing delirium in a variety of medical settings, with a pooled sensitivity of 82% and specificity of 99% when compared to the gold-standard DSM criteria in meta-analysis (Shi et al., Reference Shi, Warren, Saposnik and Macdermid2013). Although the CAM has not been studied extensively in patients with dementia, existing evidence from small studies suggests that it has adequate diagnostic accuracy in this population (Morandi et al., Reference Morandi2012).
Inouye’s chart abstraction tool (used in two of the studies identified by Gupta et al., Reference Gupta2022) has been shown to have a sensitivity of 74% and specificity of 83% using the CAM as the gold standard (Inouye et al., Reference Inouye, Leo-Summers, Zhang, Bogardus, Leslie and Agostini2005). However, dementia has been shown to be associated with false positive delirium ratings using this chart abstraction tool, with a dementia diagnosis present in 20% of the false positive ratings (vs. 9% of the true negative ratings) (Inouye et al., Reference Inouye, Leo-Summers, Zhang, Bogardus, Leslie and Agostini2005). Thus, dementia, which is a significant risk factor for delirium, has the potential to confound the outcomes of clinical trials in which the diagnosis of delirium was made via retrospective chart review.
The most straightforward potential reason for negative pharmacological studies is that existing treatments are simply not effective. Delirium is often multi-factorial, particularly in older adults with multimorbidity and polypharmacy, and a single medication is likely often not enough to address all of the contributing variables. For example, gabapentinoids may reduce opioid use, but do not address other potential peri-operative contributors to delirium, such as sleep disruption, anesthetic agents, reduced mobility, and dehydration. Alternatively, medication treatment may improve some delirium symptoms, but not enough to change patients’ overall outcome based on standardized delirium rating scales. For example, antipsychotics can potentially reduce agitation and psychosis and help to consolidate sleep, but do not necessarily improve cognitive performance and awareness (or could make these features worse). This variability in delirium symptom response might be particularly pertinent for persons with dementia who have cognitive and behavioral symptoms at baseline, making it more difficult to measure treatment effect.
The most studied medications for delirium treatment are antipsychotics, primarily haloperidol, and second-generation antipsychotics. However, despite widespread clinical use, existing evidence suggests that antipsychotics are no better than placebo in reducing delirium severity or mortality (Burry et al., Reference Burry2018). In their Cochrane review examining the use of antipsychotics to treat delirium in hospitalized non-ICU patients, Burry et al. not only identified the overall null findings but also noted the majority of the RCTs in this area to be of low methodological quality (Burry et al., Reference Burry2018), meaning that one cannot be confident in the findings. The authors identified methodological limitations such as inadequate sample sizes, attrition bias, variable outcome measures (e.g., delirium severity versus delirium resolution) and measurement instruments, and inadequate reporting of use of additional non-pharmacological strategies, sedating medications (e.g., benzodiazepines), and physical restraints. They also noted that studies did not generally report on adverse events or length of hospital stay or hospital disposition, potentially clinically important outcome variables.
Fortunately, several initiatives are underway to standardize delirium research. The Network for Investigation of Delirium: Unifying Scientists (NIDUS) (https://deliriumnetwork.org) is working to develop a core outcome set of patient-centered outcomes and evidence-based measurement instruments for use in delirium research (Rose et al., Reference Rose2017). The International Federation of Delirium Societies (iDelirium; http://www.idelirium.org/) has proposed a diagnostic framework to improve the diagnosis of delirium superimposed on dementia (Morandi et al., Reference Morandi2017). In addition to standardization of outcomes and measurement instruments in delirium clinical trials, adequately powered prospective placebo-controlled RCTs, particularly in hospitalized non-ICU patients (the most common, but least studied patient group), are needed to determine the efficacy of delirium treatment options. Further, as suggested by Burry et al., clinical trials should evaluate the impact of medication administered in the context of standardized non-pharmacological delirium prevention and treatment strategies (Burry et al., Reference Burry2018). Ideally, clinical trials would include pre-specified subgroup analyses examining the efficacy of interventions in patients with dementia, as they are both at high risk for delirium and potentially less likely to respond to treatment (and to have more adverse effects) than persons with less neurological vulnerability. Hopefully, with high-quality, interdisciplinary and collaborative research, patients with delirium will have better treatment options in the coming years.
Disclosures
K. S. Bingham reports no financial relationships with commercial interests. She currently receives research support from the University of Toronto Department of Psychiatry. A. J. Flint receives grant support from the U.S. National Institutes of Health, the Patient-Centered Outcomes Research Institute, the Canadian Institutes of Health Research, Brain Canada, AGE-WELL, and the Canadian Foundation for Healthcare Improvement.