Helen C. Kales, MD

According to the U.S. Department of Health and Human Services, in 2007, 88% of 1.4 million Medicare claims for second-generation antipsychotics (SGAs) in older adult nursing home residents were associated with a dementia diagnosis. Similar trends have been observed in Canada and Europe.^1-4 In a retrospective analysis of medication data from older residents with dementia in 6 care homes in England, long-term (ie, >1 month) use of antipsychotics was the most frequent potentially inappropriate prescribing practice.³ In another study in 7 European countries and Israel, the overall prevalence of antipsychotic use among long-term care residents with dementia was 33%.¹ Similarly, a recent literature review⁵ found that 22% to 86% of antipsychotic prescriptions to older individuals were off-label; this practice was particularly common for individuals with agitation.

Because of the aging population and widespread prescription of antipsychotics to older patients, clinicians need information on the relative risks of using these medications in this population. In the United States, all antipsychotics carry a FDA “black-box” warning of the increased risk of death in older adults with dementia. In addition, the risk of death is increased when prescribing antipsychotics to older adults with other conditions, such as Parkinson’s disease,⁶ and other safety and tolerability concerns, including falls and fractures, sedation, metabolic abnormalities, and extrapyramidal effects, are highly relevant to geriatric patients.

This 3-part review summarizes findings and recommendations on prescribing antipsychotics to older individuals with schizophrenia, bipolar disorder, depression, and dementia. This third and final installment:

• briefly summarizes the major studies and analyses relevant to prescribing antipsychotics to older patients with dementia
• provides a summative opinion on safety and tolerability issues in these older patients
• highlights the gaps in the evidence base and areas that need additional research.

Summary of benefits, place in treatment armamentarium

Behavioral and psychological symptoms of dementia (BPSD) include agitation, delusional beliefs, repetitive questioning, hallucinations, aggression, wandering, and various socially inappropriate behaviors.⁷ These occur almost universally in all types and stages of dementia.⁷ BPSD are among the most complex, stressful, and costly aspects of dementia care, and lead to a myriad of poor health outcomes, including excess morbidity, mortality, hospital stays, and early nursing home placement.^8-11 Because BPSD usually occur across all types and stages of dementia,^7,12-16 the prevalence of BPSD mirrors the overall prevalence of dementia.

Although all expert organizations, including the American Psychiatric Association,¹⁷ recommend nonpharmacologic strategies as first-line treatment for BPSD, for the most part, these recommendations have not been translated into standard clinical management or routine care.¹⁸ Because of a perceived lack of other options, the current mainstay of treatment is the off-label use of psychotropics such as antipsy­chotics. Of all the agents currently used for BPSD, SGAs have the strongest evidence base, although benefits are modest at best (standardized effect size 0.13 to 0.16).^19,20 In terms of individual SGAs, only risperidone is indicated for aggression in Canada and in Europe (not in the United States); risperidone has the best evidence for efficacy, with a meta-analysis of 5 published randomized controlled trials (RCTs) reporting that risperidone is superior to other SGAs for aggression in dementia.^21,22 As a class, first-generation antipsychotics (FGAs) have no clear evidence for BPSD as defined broadly; however, there may be slight benefit for haloperidol for aggression.^23,24

Clinical Trials

Adverse effects. A meta-analysis of RCTs of SGAs found that, compared with placebo, SGAs have increased rates of several adverse effects. These include somnolence (17% drug vs 7% placebo; odds ratio [OR], 2.84; 95% confidence interval [CI], 2.25 to 3.58; P < .00001); extrapyramidal symptoms (13% drug vs 8% placebo; OR, 1.51; 95% CI, 1.20 to 1.91; P = .0005; primarily attributable to risperidone); abnormal gait (10% drug vs 2% placebo; OR, 3.42; 95% CI, 1.78 to 6.56; P = .0002; attributable to olanzapine and risperidone); edema (9% drug vs 4% placebo; OR, 1.99; 95% CI, 1.20 to 3.30; P = .008; attributable to olanzapine and risperidone); urinary tract infections/incontinence (16% drug vs 12% placebo; OR, 1.28; 95% CI, 1.02 to 1.61; P = .04); cognitive impairment measured as difference in Mini-Mental State Examination score (95% CI, 0.38 to 1.09; P < .0001)²⁵; and stroke (1.9% drug vs 0.9% placebo, OR, 2.13; 95% CI, 1.20 to 3.75; P = .009).^21,26

In the 42-site Clinical Antipsychotic Trials of Intervention Effectiveness Alzheimer’s disease RCT, 421 outpatients with Alzheimer’s disease and BPSD were randomized to an SGA (risperidone, olanzapine, or quetiapine) or placebo. Compared with placebo, SGAs had a higher rate of parkinsonism or extrapyramidal signs (olanzapine and risperidone groups); sedation; confusion/changes in mental status (olanzapine and risperidone); psychotic symptoms (olanzapine); and increase in body weight and body mass index.²⁶

In the 2005 FDA black-box warning, pneumonia and cardiac adverse effects were cited as primary causes of death for patients with dementia taking SGAs. A subsequent observational study confirmed that use of either FGAs or SGAs in geriatric patients was associated with an increased risk of pneumonia, in a dose-dependent manner.²⁷ Although there is limited data on cardiac adverse effects in older adults, especially those with dementia taking antipsychotics,²⁸ 1 observational study of nursing home residents²⁹ found that those taking FGAs had a significantly higher risk of hospitalization for ventricular arrhythmia or cardiac arrest compared with those who were not taking FGAs. In contrast, there was no increased risk with SGAs.

Mortality.
 In 2005, the FDA announced that based on a reanalysis of 17 placebo-controlled trials (many of which were unpublished) that SGAs were associated with a 1.7-fold increase in mortality compared with placebo.³⁰ As a result, the FDA issued a black-box warning for using SGAs in patients with dementia. The overall OR in a published meta-analysis of mortality with SGAs was 1.54 (1.06 to 2.23; z = 2.28; P = .02), with pooled events of 3.5% mortality vs 2.3% (drug vs placebo).²¹ This meta-analysis²¹ also included ad hoc analyses of haloperidol; using combined data from 2 contrasts of haloperidol (with risperidone and quetiapine; 243 patients receiving haloperidol and 239 receiving placebo) they also found 15 deaths (6.2%) with haloperidol and 9 (3.8%) with placebo, resulting in an OR of 1.68.

Other clinical data

Observational studies. Most observational studies have confirmed concerns regarding increased mortality in patients with BPSD who take antipsychotics, with FGAs having a higher risk than SGAs^18,31 and SGAs having a higher risk compared with most other psychotropics.³² Three studies that found no increase in mortality with antipsychotics in patients with dementia had methodological issues, including examining prevalence as opposed to new users,^33,34 not controlling for exposure,^10,33,34 power issues,^10,34 not controlling for other psychiatric medications,¹⁰ and varying lengths of follow-up.¹⁰ An FDA black-box warning for FGAs was announced in 2008³⁰ based on 2 observational studies that showed an increased risk of mortality in older adults taking FGAs vs SGAs.^35,36

In terms of specific SGAs, Kales et al³⁷ examined the mortality risk associated with individual antipsychotics using various methods to control for confounding. Among a national sample of >33,000 older veterans with dementia newly started on haloperidol, risperidone, olanzapine, quetiapine, or valproic acid and derivatives (as a nonantipsychotic comparator), the highest mortality across all analyses (intent to treat, exposure, propensity-adjusted) was associated with haloperidol, followed by risperidone and olanzapine, valproic acid, and quetiapine.

Most recently, a retrospective case-control study (90,786 patients age ≥65 with dementia) examined the number needed to harm (NNH; ie, number of patients needed to receive treatment that would result in 1 death) over 180 days following initiation of an FGA or SGA.³⁸ This study found the following NNHs: haloperidol, 26 (95% CI, 15 to 99); risperidone, 27 (95% CI, 19 to 46); olanzapine, 40 (95% CI, 21 to 312); and quetiapine, 50 (95% CI, 30 to 150).³⁸ These results are congruent with a review of observational studies that found the highest risk of mortality was associated with haloperidol and chlorpromazine, and the lowest risk with olanzapine, quetiapine, and ziprasidone.²⁸

Patterns of antipsychotic use in older dementia patients

There are high rates of antipsychotic use in patients with dementia. Before the FDA issued the black-box warning, the Aging Demographics and Memory study found that the rate of antipsychotic use in community (outpatient) older adults with dementia was approximately 19% between 2002 and 2004 in a representative sample of 307 older adults.³⁹ Another study examining trends in community antipsychotic use in the U.S. Department of Veterans Affairs (VA) found that in the 1990s, SGA use was increasing; approximately 18% of outpatients with dementia were taking these agents.⁴⁰ Use of SGAs began to decline in 2003, ahead of the 2005 black-box warning, in tandem with other advisories (eg, diabetes, metabolic syndrome,⁴¹ and stroke risk).^42,43 Olanzapine and risperidone showed declining rates between 2003 and 2005, whereas quetiapine use significantly increased during this period. All 3 SGAs declined after the black-box warning. However, by the end of 2007, the use of SGAs had leveled off to approximately 12% of VA patients with dementia. A recent U.S. Government Accountability Office (GAO) report found that in 2012, 14% of older adult Medicare Part D enrollees with dementia living in the community were prescribed an antipsychotic.⁴⁴

Use in nursing home residents. Because BPSD are one of the main reasons people with dementia are placed in nursing homes, it is not surprising that rates of antipsychotic use are higher in these settings than in the community. Prior to the black-box warning, studies found that 24% to 32% of nursing home residents were treated with antipsychotics.^45-47 A study examining VA nursing homes (n = 133 facilities, n = 3,692 veterans) found that approximately 26% of residents were prescribed antipsychotics in 2004 to 2005.⁴⁸ The Center for Medicare and Medicaid Services (CMS) National Partnership to Improve Dementia Care in Nursing Homes has appeared to lower antipsychotic medication use in nursing homes; the rate decreased from 24% in long-stay nursing home residents nationwide in 2011 to 19% by the end of 2014. Specific to dementia, a 2010 CMS report⁴⁹ indicated that approximately 40% of nursing home residents with cognitive impairment and behavioral issues, without psychosis, received antipsychotics. The GAO data indicated that approximately 33% of older Medicare Part D enrollees with dementia who spent >100 days in a nursing home were prescribed an antipsychotic in 2012.⁴⁴ A recent Canadian study using drug claims data found that overall psychotropic use in patients with dementia remains high, finding that three-fourths of all patients with dementia in long-term care are given at least 1 psychotropic, and up to one-third are prescribed SGAs.⁵⁰ European data similarly show that antipsychotics continue to be prescribed to up to one-third of long-term care residents with dementia, with 7 out of 10 receiving an SGA.¹

Conclusions

The Table provides a summary of the evidence regarding the use of antipsychotics in patients with dementia. Expert consensus is that among BPSD, aggression and psychosis are the primary indications for using antipsychotics.⁵¹ Based on multiple RCTs and meta-analyses, the evidence for using SGAs to treat these symptoms is moderate at best. However, in real-world practice settings, SGAs are widely used for symptoms, such as wandering, inappropriate behaviors, resistance to care, etc., for which there is no evidence for efficacy other than sedation. Furthermore, even when there is a potential for benefit, this must be balanced against the risk of adverse effects, including somnolence, worsened cognition, extra­pyramidal symptoms, stroke, and mortality.
 

Clinicians who care for older adults with BPSD should strive to increase the use of first-line nonpharmacologic strategies, by using structured approaches such as DICE (Describe, Investigate, Create, Evaluate) described in the Box.⁵¹ Antipsychotics should be reserved for situations in which nonpharmacologic approaches are unsuccessful, or there is concern for serious or imminent risk to the patient or others.

In the future, observational studies using biomarkers, such as neuroimaging markers, of brain health in older patients taking antipsychotics for various durations may give us a better understanding of long-term antipsychotic safety and tolerability and the monitoring required to assess long-term burden of specific antipsychotics in real-world samples.⁵² However, because of various biases, observational data may not provide answers to all questions,⁵³ and a major challenge is that the number of published RCTs specific to geriatric patients is not growing substantially. Pharmacotherapy evidence is not keeping up with demographic trends. Key developments in RCTs will be the inclusion of biomarkers via neuroimaging, drug serum or brain levels, and genetic profiling. Because of the modest findings of benefits of antipsychotics in dementia and safety concerns addressing brain health in preclinical or early stages, identification of effective non-drug interventions and identifying true disease-modifying agents will be the next challenges of dementia research.
 

According to the U.S. Department of Health and Human Services, in 2007, 88% of 1.4 million Medicare claims for second-generation antipsychotics (SGAs) in older adult nursing home residents were associated with a dementia diagnosis. Similar trends have been observed in Canada and Europe.^1-4 In a retrospective analysis of medication data from older residents with dementia in 6 care homes in England, long-term (ie, >1 month) use of antipsychotics was the most frequent potentially inappropriate prescribing practice.³ In another study in 7 European countries and Israel, the overall prevalence of antipsychotic use among long-term care residents with dementia was 33%.¹ Similarly, a recent literature review⁵ found that 22% to 86% of antipsychotic prescriptions to older individuals were off-label; this practice was particularly common for individuals with agitation.

Because of the aging population and widespread prescription of antipsychotics to older patients, clinicians need information on the relative risks of using these medications in this population. In the United States, all antipsychotics carry a FDA “black-box” warning of the increased risk of death in older adults with dementia. In addition, the risk of death is increased when prescribing antipsychotics to older adults with other conditions, such as Parkinson’s disease,⁶ and other safety and tolerability concerns, including falls and fractures, sedation, metabolic abnormalities, and extrapyramidal effects, are highly relevant to geriatric patients.

This 3-part review summarizes findings and recommendations on prescribing antipsychotics to older individuals with schizophrenia, bipolar disorder, depression, and dementia. This third and final installment:

• briefly summarizes the major studies and analyses relevant to prescribing antipsychotics to older patients with dementia
• provides a summative opinion on safety and tolerability issues in these older patients
• highlights the gaps in the evidence base and areas that need additional research.

Summary of benefits, place in treatment armamentarium

Behavioral and psychological symptoms of dementia (BPSD) include agitation, delusional beliefs, repetitive questioning, hallucinations, aggression, wandering, and various socially inappropriate behaviors.⁷ These occur almost universally in all types and stages of dementia.⁷ BPSD are among the most complex, stressful, and costly aspects of dementia care, and lead to a myriad of poor health outcomes, including excess morbidity, mortality, hospital stays, and early nursing home placement.^8-11 Because BPSD usually occur across all types and stages of dementia,^7,12-16 the prevalence of BPSD mirrors the overall prevalence of dementia.

Although all expert organizations, including the American Psychiatric Association,¹⁷ recommend nonpharmacologic strategies as first-line treatment for BPSD, for the most part, these recommendations have not been translated into standard clinical management or routine care.¹⁸ Because of a perceived lack of other options, the current mainstay of treatment is the off-label use of psychotropics such as antipsy­chotics. Of all the agents currently used for BPSD, SGAs have the strongest evidence base, although benefits are modest at best (standardized effect size 0.13 to 0.16).^19,20 In terms of individual SGAs, only risperidone is indicated for aggression in Canada and in Europe (not in the United States); risperidone has the best evidence for efficacy, with a meta-analysis of 5 published randomized controlled trials (RCTs) reporting that risperidone is superior to other SGAs for aggression in dementia.^21,22 As a class, first-generation antipsychotics (FGAs) have no clear evidence for BPSD as defined broadly; however, there may be slight benefit for haloperidol for aggression.^23,24

Clinical Trials

Adverse effects. A meta-analysis of RCTs of SGAs found that, compared with placebo, SGAs have increased rates of several adverse effects. These include somnolence (17% drug vs 7% placebo; odds ratio [OR], 2.84; 95% confidence interval [CI], 2.25 to 3.58; P < .00001); extrapyramidal symptoms (13% drug vs 8% placebo; OR, 1.51; 95% CI, 1.20 to 1.91; P = .0005; primarily attributable to risperidone); abnormal gait (10% drug vs 2% placebo; OR, 3.42; 95% CI, 1.78 to 6.56; P = .0002; attributable to olanzapine and risperidone); edema (9% drug vs 4% placebo; OR, 1.99; 95% CI, 1.20 to 3.30; P = .008; attributable to olanzapine and risperidone); urinary tract infections/incontinence (16% drug vs 12% placebo; OR, 1.28; 95% CI, 1.02 to 1.61; P = .04); cognitive impairment measured as difference in Mini-Mental State Examination score (95% CI, 0.38 to 1.09; P < .0001)²⁵; and stroke (1.9% drug vs 0.9% placebo, OR, 2.13; 95% CI, 1.20 to 3.75; P = .009).^21,26

In the 42-site Clinical Antipsychotic Trials of Intervention Effectiveness Alzheimer’s disease RCT, 421 outpatients with Alzheimer’s disease and BPSD were randomized to an SGA (risperidone, olanzapine, or quetiapine) or placebo. Compared with placebo, SGAs had a higher rate of parkinsonism or extrapyramidal signs (olanzapine and risperidone groups); sedation; confusion/changes in mental status (olanzapine and risperidone); psychotic symptoms (olanzapine); and increase in body weight and body mass index.²⁶

In the 2005 FDA black-box warning, pneumonia and cardiac adverse effects were cited as primary causes of death for patients with dementia taking SGAs. A subsequent observational study confirmed that use of either FGAs or SGAs in geriatric patients was associated with an increased risk of pneumonia, in a dose-dependent manner.²⁷ Although there is limited data on cardiac adverse effects in older adults, especially those with dementia taking antipsychotics,²⁸ 1 observational study of nursing home residents²⁹ found that those taking FGAs had a significantly higher risk of hospitalization for ventricular arrhythmia or cardiac arrest compared with those who were not taking FGAs. In contrast, there was no increased risk with SGAs.

Mortality.
 In 2005, the FDA announced that based on a reanalysis of 17 placebo-controlled trials (many of which were unpublished) that SGAs were associated with a 1.7-fold increase in mortality compared with placebo.³⁰ As a result, the FDA issued a black-box warning for using SGAs in patients with dementia. The overall OR in a published meta-analysis of mortality with SGAs was 1.54 (1.06 to 2.23; z = 2.28; P = .02), with pooled events of 3.5% mortality vs 2.3% (drug vs placebo).²¹ This meta-analysis²¹ also included ad hoc analyses of haloperidol; using combined data from 2 contrasts of haloperidol (with risperidone and quetiapine; 243 patients receiving haloperidol and 239 receiving placebo) they also found 15 deaths (6.2%) with haloperidol and 9 (3.8%) with placebo, resulting in an OR of 1.68.

Other clinical data

Observational studies. Most observational studies have confirmed concerns regarding increased mortality in patients with BPSD who take antipsychotics, with FGAs having a higher risk than SGAs^18,31 and SGAs having a higher risk compared with most other psychotropics.³² Three studies that found no increase in mortality with antipsychotics in patients with dementia had methodological issues, including examining prevalence as opposed to new users,^33,34 not controlling for exposure,^10,33,34 power issues,^10,34 not controlling for other psychiatric medications,¹⁰ and varying lengths of follow-up.¹⁰ An FDA black-box warning for FGAs was announced in 2008³⁰ based on 2 observational studies that showed an increased risk of mortality in older adults taking FGAs vs SGAs.^35,36

In terms of specific SGAs, Kales et al³⁷ examined the mortality risk associated with individual antipsychotics using various methods to control for confounding. Among a national sample of >33,000 older veterans with dementia newly started on haloperidol, risperidone, olanzapine, quetiapine, or valproic acid and derivatives (as a nonantipsychotic comparator), the highest mortality across all analyses (intent to treat, exposure, propensity-adjusted) was associated with haloperidol, followed by risperidone and olanzapine, valproic acid, and quetiapine.

Most recently, a retrospective case-control study (90,786 patients age ≥65 with dementia) examined the number needed to harm (NNH; ie, number of patients needed to receive treatment that would result in 1 death) over 180 days following initiation of an FGA or SGA.³⁸ This study found the following NNHs: haloperidol, 26 (95% CI, 15 to 99); risperidone, 27 (95% CI, 19 to 46); olanzapine, 40 (95% CI, 21 to 312); and quetiapine, 50 (95% CI, 30 to 150).³⁸ These results are congruent with a review of observational studies that found the highest risk of mortality was associated with haloperidol and chlorpromazine, and the lowest risk with olanzapine, quetiapine, and ziprasidone.²⁸

Patterns of antipsychotic use in older dementia patients

There are high rates of antipsychotic use in patients with dementia. Before the FDA issued the black-box warning, the Aging Demographics and Memory study found that the rate of antipsychotic use in community (outpatient) older adults with dementia was approximately 19% between 2002 and 2004 in a representative sample of 307 older adults.³⁹ Another study examining trends in community antipsychotic use in the U.S. Department of Veterans Affairs (VA) found that in the 1990s, SGA use was increasing; approximately 18% of outpatients with dementia were taking these agents.⁴⁰ Use of SGAs began to decline in 2003, ahead of the 2005 black-box warning, in tandem with other advisories (eg, diabetes, metabolic syndrome,⁴¹ and stroke risk).^42,43 Olanzapine and risperidone showed declining rates between 2003 and 2005, whereas quetiapine use significantly increased during this period. All 3 SGAs declined after the black-box warning. However, by the end of 2007, the use of SGAs had leveled off to approximately 12% of VA patients with dementia. A recent U.S. Government Accountability Office (GAO) report found that in 2012, 14% of older adult Medicare Part D enrollees with dementia living in the community were prescribed an antipsychotic.⁴⁴

Use in nursing home residents. Because BPSD are one of the main reasons people with dementia are placed in nursing homes, it is not surprising that rates of antipsychotic use are higher in these settings than in the community. Prior to the black-box warning, studies found that 24% to 32% of nursing home residents were treated with antipsychotics.^45-47 A study examining VA nursing homes (n = 133 facilities, n = 3,692 veterans) found that approximately 26% of residents were prescribed antipsychotics in 2004 to 2005.⁴⁸ The Center for Medicare and Medicaid Services (CMS) National Partnership to Improve Dementia Care in Nursing Homes has appeared to lower antipsychotic medication use in nursing homes; the rate decreased from 24% in long-stay nursing home residents nationwide in 2011 to 19% by the end of 2014. Specific to dementia, a 2010 CMS report⁴⁹ indicated that approximately 40% of nursing home residents with cognitive impairment and behavioral issues, without psychosis, received antipsychotics. The GAO data indicated that approximately 33% of older Medicare Part D enrollees with dementia who spent >100 days in a nursing home were prescribed an antipsychotic in 2012.⁴⁴ A recent Canadian study using drug claims data found that overall psychotropic use in patients with dementia remains high, finding that three-fourths of all patients with dementia in long-term care are given at least 1 psychotropic, and up to one-third are prescribed SGAs.⁵⁰ European data similarly show that antipsychotics continue to be prescribed to up to one-third of long-term care residents with dementia, with 7 out of 10 receiving an SGA.¹

Conclusions

The Table provides a summary of the evidence regarding the use of antipsychotics in patients with dementia. Expert consensus is that among BPSD, aggression and psychosis are the primary indications for using antipsychotics.⁵¹ Based on multiple RCTs and meta-analyses, the evidence for using SGAs to treat these symptoms is moderate at best. However, in real-world practice settings, SGAs are widely used for symptoms, such as wandering, inappropriate behaviors, resistance to care, etc., for which there is no evidence for efficacy other than sedation. Furthermore, even when there is a potential for benefit, this must be balanced against the risk of adverse effects, including somnolence, worsened cognition, extra­pyramidal symptoms, stroke, and mortality.
 

Clinicians who care for older adults with BPSD should strive to increase the use of first-line nonpharmacologic strategies, by using structured approaches such as DICE (Describe, Investigate, Create, Evaluate) described in the Box.⁵¹ Antipsychotics should be reserved for situations in which nonpharmacologic approaches are unsuccessful, or there is concern for serious or imminent risk to the patient or others.

In the future, observational studies using biomarkers, such as neuroimaging markers, of brain health in older patients taking antipsychotics for various durations may give us a better understanding of long-term antipsychotic safety and tolerability and the monitoring required to assess long-term burden of specific antipsychotics in real-world samples.⁵² However, because of various biases, observational data may not provide answers to all questions,⁵³ and a major challenge is that the number of published RCTs specific to geriatric patients is not growing substantially. Pharmacotherapy evidence is not keeping up with demographic trends. Key developments in RCTs will be the inclusion of biomarkers via neuroimaging, drug serum or brain levels, and genetic profiling. Because of the modest findings of benefits of antipsychotics in dementia and safety concerns addressing brain health in preclinical or early stages, identification of effective non-drug interventions and identifying true disease-modifying agents will be the next challenges of dementia research.
 

According to the U.S. Department of Health and Human Services, in 2007, 88% of 1.4 million Medicare claims for second-generation antipsychotics (SGAs) in older adult nursing home residents were associated with a dementia diagnosis. Similar trends have been observed in Canada and Europe.^1-4 In a retrospective analysis of medication data from older residents with dementia in 6 care homes in England, long-term (ie, >1 month) use of antipsychotics was the most frequent potentially inappropriate prescribing practice.³ In another study in 7 European countries and Israel, the overall prevalence of antipsychotic use among long-term care residents with dementia was 33%.¹ Similarly, a recent literature review⁵ found that 22% to 86% of antipsychotic prescriptions to older individuals were off-label; this practice was particularly common for individuals with agitation.

Because of the aging population and widespread prescription of antipsychotics to older patients, clinicians need information on the relative risks of using these medications in this population. In the United States, all antipsychotics carry a FDA “black-box” warning of the increased risk of death in older adults with dementia. In addition, the risk of death is increased when prescribing antipsychotics to older adults with other conditions, such as Parkinson’s disease,⁶ and other safety and tolerability concerns, including falls and fractures, sedation, metabolic abnormalities, and extrapyramidal effects, are highly relevant to geriatric patients.

This 3-part review summarizes findings and recommendations on prescribing antipsychotics to older individuals with schizophrenia, bipolar disorder, depression, and dementia. This third and final installment:

• briefly summarizes the major studies and analyses relevant to prescribing antipsychotics to older patients with dementia
• provides a summative opinion on safety and tolerability issues in these older patients
• highlights the gaps in the evidence base and areas that need additional research.

Summary of benefits, place in treatment armamentarium

Behavioral and psychological symptoms of dementia (BPSD) include agitation, delusional beliefs, repetitive questioning, hallucinations, aggression, wandering, and various socially inappropriate behaviors.⁷ These occur almost universally in all types and stages of dementia.⁷ BPSD are among the most complex, stressful, and costly aspects of dementia care, and lead to a myriad of poor health outcomes, including excess morbidity, mortality, hospital stays, and early nursing home placement.^8-11 Because BPSD usually occur across all types and stages of dementia,^7,12-16 the prevalence of BPSD mirrors the overall prevalence of dementia.

Although all expert organizations, including the American Psychiatric Association,¹⁷ recommend nonpharmacologic strategies as first-line treatment for BPSD, for the most part, these recommendations have not been translated into standard clinical management or routine care.¹⁸ Because of a perceived lack of other options, the current mainstay of treatment is the off-label use of psychotropics such as antipsy­chotics. Of all the agents currently used for BPSD, SGAs have the strongest evidence base, although benefits are modest at best (standardized effect size 0.13 to 0.16).^19,20 In terms of individual SGAs, only risperidone is indicated for aggression in Canada and in Europe (not in the United States); risperidone has the best evidence for efficacy, with a meta-analysis of 5 published randomized controlled trials (RCTs) reporting that risperidone is superior to other SGAs for aggression in dementia.^21,22 As a class, first-generation antipsychotics (FGAs) have no clear evidence for BPSD as defined broadly; however, there may be slight benefit for haloperidol for aggression.^23,24

Clinical Trials

Adverse effects. A meta-analysis of RCTs of SGAs found that, compared with placebo, SGAs have increased rates of several adverse effects. These include somnolence (17% drug vs 7% placebo; odds ratio [OR], 2.84; 95% confidence interval [CI], 2.25 to 3.58; P < .00001); extrapyramidal symptoms (13% drug vs 8% placebo; OR, 1.51; 95% CI, 1.20 to 1.91; P = .0005; primarily attributable to risperidone); abnormal gait (10% drug vs 2% placebo; OR, 3.42; 95% CI, 1.78 to 6.56; P = .0002; attributable to olanzapine and risperidone); edema (9% drug vs 4% placebo; OR, 1.99; 95% CI, 1.20 to 3.30; P = .008; attributable to olanzapine and risperidone); urinary tract infections/incontinence (16% drug vs 12% placebo; OR, 1.28; 95% CI, 1.02 to 1.61; P = .04); cognitive impairment measured as difference in Mini-Mental State Examination score (95% CI, 0.38 to 1.09; P < .0001)²⁵; and stroke (1.9% drug vs 0.9% placebo, OR, 2.13; 95% CI, 1.20 to 3.75; P = .009).^21,26

In the 42-site Clinical Antipsychotic Trials of Intervention Effectiveness Alzheimer’s disease RCT, 421 outpatients with Alzheimer’s disease and BPSD were randomized to an SGA (risperidone, olanzapine, or quetiapine) or placebo. Compared with placebo, SGAs had a higher rate of parkinsonism or extrapyramidal signs (olanzapine and risperidone groups); sedation; confusion/changes in mental status (olanzapine and risperidone); psychotic symptoms (olanzapine); and increase in body weight and body mass index.²⁶

In the 2005 FDA black-box warning, pneumonia and cardiac adverse effects were cited as primary causes of death for patients with dementia taking SGAs. A subsequent observational study confirmed that use of either FGAs or SGAs in geriatric patients was associated with an increased risk of pneumonia, in a dose-dependent manner.²⁷ Although there is limited data on cardiac adverse effects in older adults, especially those with dementia taking antipsychotics,²⁸ 1 observational study of nursing home residents²⁹ found that those taking FGAs had a significantly higher risk of hospitalization for ventricular arrhythmia or cardiac arrest compared with those who were not taking FGAs. In contrast, there was no increased risk with SGAs.

Mortality.
 In 2005, the FDA announced that based on a reanalysis of 17 placebo-controlled trials (many of which were unpublished) that SGAs were associated with a 1.7-fold increase in mortality compared with placebo.³⁰ As a result, the FDA issued a black-box warning for using SGAs in patients with dementia. The overall OR in a published meta-analysis of mortality with SGAs was 1.54 (1.06 to 2.23; z = 2.28; P = .02), with pooled events of 3.5% mortality vs 2.3% (drug vs placebo).²¹ This meta-analysis²¹ also included ad hoc analyses of haloperidol; using combined data from 2 contrasts of haloperidol (with risperidone and quetiapine; 243 patients receiving haloperidol and 239 receiving placebo) they also found 15 deaths (6.2%) with haloperidol and 9 (3.8%) with placebo, resulting in an OR of 1.68.

Other clinical data

Observational studies. Most observational studies have confirmed concerns regarding increased mortality in patients with BPSD who take antipsychotics, with FGAs having a higher risk than SGAs^18,31 and SGAs having a higher risk compared with most other psychotropics.³² Three studies that found no increase in mortality with antipsychotics in patients with dementia had methodological issues, including examining prevalence as opposed to new users,^33,34 not controlling for exposure,^10,33,34 power issues,^10,34 not controlling for other psychiatric medications,¹⁰ and varying lengths of follow-up.¹⁰ An FDA black-box warning for FGAs was announced in 2008³⁰ based on 2 observational studies that showed an increased risk of mortality in older adults taking FGAs vs SGAs.^35,36

In terms of specific SGAs, Kales et al³⁷ examined the mortality risk associated with individual antipsychotics using various methods to control for confounding. Among a national sample of >33,000 older veterans with dementia newly started on haloperidol, risperidone, olanzapine, quetiapine, or valproic acid and derivatives (as a nonantipsychotic comparator), the highest mortality across all analyses (intent to treat, exposure, propensity-adjusted) was associated with haloperidol, followed by risperidone and olanzapine, valproic acid, and quetiapine.

Most recently, a retrospective case-control study (90,786 patients age ≥65 with dementia) examined the number needed to harm (NNH; ie, number of patients needed to receive treatment that would result in 1 death) over 180 days following initiation of an FGA or SGA.³⁸ This study found the following NNHs: haloperidol, 26 (95% CI, 15 to 99); risperidone, 27 (95% CI, 19 to 46); olanzapine, 40 (95% CI, 21 to 312); and quetiapine, 50 (95% CI, 30 to 150).³⁸ These results are congruent with a review of observational studies that found the highest risk of mortality was associated with haloperidol and chlorpromazine, and the lowest risk with olanzapine, quetiapine, and ziprasidone.²⁸

Patterns of antipsychotic use in older dementia patients

There are high rates of antipsychotic use in patients with dementia. Before the FDA issued the black-box warning, the Aging Demographics and Memory study found that the rate of antipsychotic use in community (outpatient) older adults with dementia was approximately 19% between 2002 and 2004 in a representative sample of 307 older adults.³⁹ Another study examining trends in community antipsychotic use in the U.S. Department of Veterans Affairs (VA) found that in the 1990s, SGA use was increasing; approximately 18% of outpatients with dementia were taking these agents.⁴⁰ Use of SGAs began to decline in 2003, ahead of the 2005 black-box warning, in tandem with other advisories (eg, diabetes, metabolic syndrome,⁴¹ and stroke risk).^42,43 Olanzapine and risperidone showed declining rates between 2003 and 2005, whereas quetiapine use significantly increased during this period. All 3 SGAs declined after the black-box warning. However, by the end of 2007, the use of SGAs had leveled off to approximately 12% of VA patients with dementia. A recent U.S. Government Accountability Office (GAO) report found that in 2012, 14% of older adult Medicare Part D enrollees with dementia living in the community were prescribed an antipsychotic.⁴⁴

Use in nursing home residents. Because BPSD are one of the main reasons people with dementia are placed in nursing homes, it is not surprising that rates of antipsychotic use are higher in these settings than in the community. Prior to the black-box warning, studies found that 24% to 32% of nursing home residents were treated with antipsychotics.^45-47 A study examining VA nursing homes (n = 133 facilities, n = 3,692 veterans) found that approximately 26% of residents were prescribed antipsychotics in 2004 to 2005.⁴⁸ The Center for Medicare and Medicaid Services (CMS) National Partnership to Improve Dementia Care in Nursing Homes has appeared to lower antipsychotic medication use in nursing homes; the rate decreased from 24% in long-stay nursing home residents nationwide in 2011 to 19% by the end of 2014. Specific to dementia, a 2010 CMS report⁴⁹ indicated that approximately 40% of nursing home residents with cognitive impairment and behavioral issues, without psychosis, received antipsychotics. The GAO data indicated that approximately 33% of older Medicare Part D enrollees with dementia who spent >100 days in a nursing home were prescribed an antipsychotic in 2012.⁴⁴ A recent Canadian study using drug claims data found that overall psychotropic use in patients with dementia remains high, finding that three-fourths of all patients with dementia in long-term care are given at least 1 psychotropic, and up to one-third are prescribed SGAs.⁵⁰ European data similarly show that antipsychotics continue to be prescribed to up to one-third of long-term care residents with dementia, with 7 out of 10 receiving an SGA.¹

Conclusions

The Table provides a summary of the evidence regarding the use of antipsychotics in patients with dementia. Expert consensus is that among BPSD, aggression and psychosis are the primary indications for using antipsychotics.⁵¹ Based on multiple RCTs and meta-analyses, the evidence for using SGAs to treat these symptoms is moderate at best. However, in real-world practice settings, SGAs are widely used for symptoms, such as wandering, inappropriate behaviors, resistance to care, etc., for which there is no evidence for efficacy other than sedation. Furthermore, even when there is a potential for benefit, this must be balanced against the risk of adverse effects, including somnolence, worsened cognition, extra­pyramidal symptoms, stroke, and mortality.
 

Clinicians who care for older adults with BPSD should strive to increase the use of first-line nonpharmacologic strategies, by using structured approaches such as DICE (Describe, Investigate, Create, Evaluate) described in the Box.⁵¹ Antipsychotics should be reserved for situations in which nonpharmacologic approaches are unsuccessful, or there is concern for serious or imminent risk to the patient or others.

In the future, observational studies using biomarkers, such as neuroimaging markers, of brain health in older patients taking antipsychotics for various durations may give us a better understanding of long-term antipsychotic safety and tolerability and the monitoring required to assess long-term burden of specific antipsychotics in real-world samples.⁵² However, because of various biases, observational data may not provide answers to all questions,⁵³ and a major challenge is that the number of published RCTs specific to geriatric patients is not growing substantially. Pharmacotherapy evidence is not keeping up with demographic trends. Key developments in RCTs will be the inclusion of biomarkers via neuroimaging, drug serum or brain levels, and genetic profiling. Because of the modest findings of benefits of antipsychotics in dementia and safety concerns addressing brain health in preclinical or early stages, identification of effective non-drug interventions and identifying true disease-modifying agents will be the next challenges of dementia research.
 

The proportion of older adults in the world population is growing rapidly. In the next 10 to 15 years, the population age >60 will grow 3.5 times more rapidly than the general population.¹ As a result, there is an increased urgency in examining benefits vs risks of antipsychotics in older individuals. In a 2010 U.S. nationally representative observational study, antipsychotic use was observed to rise slowly during early and middle adulthood, peaking at approximately age 55, declining slightly between ages 55 and 65, and then rising again after age 65, with >2% of individuals ages 80 to 84 receiving an anti­psychotic.² This is likely due to the chronology of psychotic, mood, and neurocognitive disorders across the life span. In this large national study, long-term antipsychotic treatment was common, and older patients were more likely to receive their prescriptions from non-psychiatrist physicians than from psychiatrists.² Among patients receiving an antipsychotic, the proportion of those receiving it for >120 days was 54% for individuals ages 70 to 74; 49% for individuals ages 75 to 79; and 46% for individuals ages 80 to 84.

This 3-part review summarizes findings and risk–benefit considerations when prescribing antipsychotics to older individuals. Part 1 focused on those with chronic psychotic disorders, such as schizophrenia or bipolar disorder,³ and part 3 will cover patients with dementia. This review (part 2) aims to:

• briefly summarize the results of randomized controlled trials (RCTs) of second-generation antipsychotics (SGAs) and other major studies and analyses in older patients with major depressive disorder (MDD)
• provide a summative opinion on the relative risks and benefits associated with using antipsychotics in older adults with MDD
• highlight the gaps in the evidence base and areas that need additional research.

Summary of benefits, place in treatment armamentarium

The prevalence of MDD and clinically significant depressive symptoms in community­dwelling older adults is 3% to 4% and 15%, respectively, and as high as 16% and 50%, respectively, in nursing home residents.⁴ Because late-life depression is associated with suffering, disability, and excessive mortality, it needs to be recognized and treated aggressively.⁵ Antidepressants are the mainstay of pharmacotherapy for late-life depression. Guidelines and expert opinion informed by the current evidence recommend using selective serotonin reuptake inhibitors, such as escitalopram or sertraline, as a first-line treatment; serotonin norepinephrine reuptake inhibitors, such as duloxetine or venlafaxine, as a second-line treatment; and other antidepressants, such as bupropion or nortriptyline, as a third-line treatment.^5,6 However, antipsychotics also have a role in treating late-life depression.

Over the past decade, several anti­psychotics have been FDA-approved for treating MDD: aripiprazole and brexpiprazole as adjunctive treatment of MDD in adults; olanzapine-fluoxetine combination for acute and maintenance treatment of treatment-resistant depression in adults and geriatric adults; and quetiapine extended-release (XR) as monotherapy for MDD in adults and as adjunctive treatment of MDD in adults and geriatric adults who have had an inadequate response to antidepressants alone (Table 1). However, “black-box” warnings for all first-generation antipsychotics (FGAs) and SGAs alert clinicians that these medications have been associated with serious adverse events in older adults with dementia, including “deaths […] due to heart-related events (eg, heart failure, sudden death) or infections (mostly pneumonia),” with 15 of 17 placebo-controlled trials showing a higher number of deaths with an antipsychotic compared with placebo.⁷ Although similar controlled data on the mortality risk of antipsychotics in older adults with mood disorders do not exist, most experts limit their use to 2 groups of older patients: those with MDD and psychotic features (“psychotic depression”) and those with treatment-resistant depression.

Clinical trials

Olanzapine plus sertraline as first-line pharmacotherapy for MDD with psychotic features. Meyers et al¹¹ reported on a double-blind randomized comparison of olanzapine plus placebo vs olanzapine plus sertraline in 259 patients with MDD with psychotic features. An unusual feature of this trial is that it included a similar number of younger and older participants (ages 18 to 93): 117 participants were age <60 (mean age [standard deviation (SD)]: 41.3 [10.8]) and 142 were age ≥60 (mean age [SD]: 71.7 [7.8]). The same dose titration schedules based on efficacy and tolerability were used in both younger and older participants. At the end of the study, the mean dose (SD) of sertraline (or placebo) did not differ significantly in younger (174.3 mg/d [34.1]) and older participants (165.7 mg/d [43.4]). However, the mean dose (SD) of olanzapine was significantly higher in younger patients (15.7 mg/d [4.7]) than in older participants (13.4 mg/d [5.1]).

In both age groups, olanzapine plus sertraline was more efficacious than olanzapine plus placebo, and there was no statistical interaction between age, time, and treatment group (ie, the trajectories of improvement were similar in older and younger patients receiving either olanzapine or olanzapine plus sertraline). Similarly, drop-out rates because of poor tolerability did not differ significantly in younger (4.3%) and older participants (5.6%). However, in a multinomial regression, older participants were more likely to discontinue treatment because of poor tolerability.²² Older participants were significantly less likely to experience weight gain (mean [SD]: +3.3 [4.9] vs +6.5 [6.6] kg) or an increase in fasting glucose and more likely to experience a fall, pedal edema, or extrapyramidal symptoms.^11,22-24 Cholesterol and triglyceride increased significantly and similarly in both age groups. The incidence of symptoms of tardive dyskinesia (TD) over the 12-week trial was low (<5%) in both younger and older participants, and clinically diagnosed TD was reported in only 1 (older) participant.²⁵

Venlafaxine plus aripiprazole for treatment-resistant MDD. In the largest double-blind randomized study of augmentation pharmacotherapy for late-life treatment-resistant depression published to date, Lenze et al²¹ compared venlafaxine plus aripiprazole vs venlafaxine plus placebo in 181 patients age >60 (mean age 66, with 49 participants age >70) with MDD who did not remit after 12 weeks of treatment with venlafaxine (up to 300 mg/d). After 12 weeks of augmentation, remission rates were significantly higher with aripiprazole than with placebo: 40 (44%) vs 26 (29%); odds ratio (95% confidence interval [CI]): 2.0 (1.1 to 3.7). The median final aripiprazole dose was 7 mg/d (range 2 to 15 mg/d) in remitters and 10 mg/d (range 2 to 15 mg/d) in nonremitters.

Five of 90 participants (5%) discontinued aripiprazole (1 each: suicide, jitteriness/akathisia, worsening parkinsonism; and 2 withdrew consent); 8 of 90 (9%) discontinued placebo (2 each: lack of efficacy, headache; 1: worsening parkinsonism; and 3 withdrew consent). The completed suicide occurred after 5 weeks of treatment with aripiprazole and was judged to be “neither due to emergent suicidal ideation nor to aripiprazole side-effects, but was concluded by investigators to be a result of the individual’s persisting and long-standing suicidal ideation.”²¹ Including the suicide, there were 4 serious adverse events (5%) in those receiving aripiprazole (1 each: suicide, congestive heart failure, mild stroke, and diverticulitis) and 2 (2%) in those receiving placebo (1 each: myocardial infarction, hospitalized for vomiting due to accidentally taking extra venlafaxine). In 86 participants receiving aripiprazole and 87 receiving placebo, the most frequently reported adverse effects were increased dream activity (aripiprazole: 23 [27%] vs placebo: 12 [14%]), weight gain (17 [20%] vs 8 [9%]), and tremor (5 [6%] vs 0). Akathisia and parkinsonism were observed more frequently with aripiprazole than with placebo (akathisia: 24 [26%] of 91 vs 11 [12%] of 90; parkinsonism: 15 [17%] of 86 vs 2 [2%] of 81). Akathisia was generally mild and resolved with dose adjustment; however, it was associated with a transient increase in suicidality in 3 (3%) participants receiving aripiprazole vs 0 receiving placebo and persisted at the end of the trial in 5 (5%) participants receiving aripiprazole vs 2 (2%) receiving placebo. Participants receiving aripiprazole had a significantly larger increase in weight (mean [SD]: +1.93 [3.00] vs +0.01 [3.15] kg), but there were no differences between aripiprazole and placebo in changes in body fat, total cholesterol, high-density lipoprotein, low-density lipoprotein, triglycerides, glucose, insulin concentration, or QTc.

Citalopram plus risperidone for treatment-resistant MDD. Alexopoulos et al²⁶ reported an analysis of data from 110 patients age ≥55 years (mean age [SD]: 63.4 [4.8]), among 489 mixed-age patients with MDD. Participants (n = 110) who did not respond to 1 to 3 antidepressants (venlafaxine, sertraline, mirtazapine, fluoxetine, paroxetine, or bupropion in >90%) during their current depressive episode completed 4 to 6 weeks of treatment with citalopram up to 40 mg/d; 93 did not respond and were treated with open-label risperidone (0.25 to 1 mg/d) augmentation for 4 to 6 weeks. Sixty-three (68%) of these 93 patients remitted and were randomized to 24 weeks of double-blind continuation treatment with citalopram plus risperidone vs citalopram plus placebo. Neither the median times to relapse (105 vs 57 days) nor the relapse rates (risperidone: 18 of 32 [56%] vs placebo: 20 of 31 [65%]) differed significantly. During the open-label risperidone augmentation, the most common adverse events were dizziness and dry mouth (n = 9 each, 9.7% of 93). During the continuation phase, headache (n = 3; 9.1% of 32) was observed with risperidone but not with placebo (n = 0). There was no incident parkinsonism or abnormal movements noted, but risperidone was associated with weight gain during both the open-label risperidone augmentation phase (mean [SD]: +0.9 [2.1] kg) and the continuation phase (risperidone: +0.8 [3.5] vs placebo: −0.3 [2.8] kg).

Quetiapine XR monotherapy for MDD. Katila et al²⁷ reported on a placebo-controlled RCT of quetiapine XR (median dose, 158.7 mg/d; range, 50 to 300 mg/d) in 338 patients age ≥66 years (mean age [SD], 71.3 [7.5]) presenting with MDD and a major depressive episode with a duration <1 year and no history of failed antidepressants trials from 2 classes (more than two-thirds of participants had not received treatment). After 9 weeks, the reduction in depressive symptoms on the Montgomery-Åsberg Depression Rating Scale was significantly larger with quetiapine XR than with placebo (mean [SD]: −16.0 [9.3] vs −9.0 [9.9]). There were congruent, significant differences between quetiapine and placebo in terms of response rate (quetiapine XR: 105 of 164 [64%] vs placebo: 52 of 171 [30.4%]) and remission rate (92 of 164 [56.1%] vs placebo: 40 of 171 [23.4%]). The drop-out rates for all causes were similar, but the drop-out rate attributed to adverse events was higher with quetiapine than placebo (16 of 166 [9.6%] vs 7 of 172 [4.1%]). Most quetiapine drop-outs were attributable to dizziness, headache, and somnolence (n = 4 each), and placebo drop-outs were because of headache (n = 2). Consistent with the profile of quetiapine, adverse events with a rate that was at least 5% higher with quetiapine than with placebo included somnolence (64 of 166 [38.6%] vs 16 of 172 [9.3%]), dry mouth (34 [20.5%] vs 18 [10.5%]), and extrapyramidal symptoms (12 [7.2%] vs 4 [2.3%]). Changes in weight and laboratory test results (eg, glucose, lipid profile) were minimal and not clinically meaningful.

Other clinical data. The efficacy and relatively good tolerability of aripiprazole in older patients with treatment-resistant depression observed in the RCT by Lenze et al²¹ is congruent with the earlier results of 2 small (N = 20 and 24) pilot studies.^18,19 In both studies, the remission rate was 50%, and the most prevalent adverse effects were agitation/restlessness/akathisia or drowsiness/sedation. Similarly, in a post hoc pooled analysis of 409 participants ages 50 to 67 from 3 placebo-controlled randomized trials, the remission rate was significantly higher with aripiprazole than with placebo (32.5% vs 17.1%), and the most common adverse effects were akathisia or restlessness (64 of 210 [30.4%]), somnolence (18 [8.6%]), and insomnia (17 [8.1%]).²⁰

Clinical considerations

When assessing the relative benefits and risks of antipsychotics in older patients, it is important to remember that conclusions and summative opinions are necessarily influenced by the source of the data. Because much of what we know about the use of antipsychotics in geriatric adults is from clinical trials, we know more about their acute efficacy and tolerability than their long-term effectiveness and safety.²⁸ There are similar issues regarding the role of antipsychotics in treating MDD in late life. Based on the results of several RCTs,^8,11 a combination of an antidepressant plus an antipsychotic is the recommended pharmacotherapy for the acute treatment of MDD with psychotic features (Table 2).^{8,11,12,19-21,23-27} However, there are no published data to guide how long the antipsychotic should be continued.²⁹

In older patients with MDD without psychotic features, 1 relatively large placebo-controlled RCT,²¹ 2 smaller open studies,^18,19 and a post hoc analysis of a large placebo-controlled RCT in mixed-age adults²⁰ support the efficacy and relatively good tolerability of aripiprazole augmentation of an antidepressant for treatment-resistant MDD. Similarly, 1 large placebo-controlled RCT supports the efficacy and relatively good tolerability of quetiapine for non–treatment-resistant MDD. However, there are no comparative data assessing the relative merits of using these antipsy­chotics vs other pharmacologic strategies (eg, switching to another antidepressant, lithium augmentation, or combination of 2 antidepressants). Because older patients are more likely to experience adverse effects that may have more serious consequences (Table 3), many prudent clinicians reserve using antipsychotics as a third-line treatment in older patients with MDD without psychotic features and limit the duration of their use to a few months.³⁰

The proportion of older adults in the world population is growing rapidly. In the next 10 to 15 years, the population age >60 will grow 3.5 times more rapidly than the general population.¹ As a result, there is an increased urgency in examining benefits vs risks of antipsychotics in older individuals. In a 2010 U.S. nationally representative observational study, antipsychotic use was observed to rise slowly during early and middle adulthood, peaking at approximately age 55, declining slightly between ages 55 and 65, and then rising again after age 65, with >2% of individuals ages 80 to 84 receiving an anti­psychotic.² This is likely due to the chronology of psychotic, mood, and neurocognitive disorders across the life span. In this large national study, long-term antipsychotic treatment was common, and older patients were more likely to receive their prescriptions from non-psychiatrist physicians than from psychiatrists.² Among patients receiving an antipsychotic, the proportion of those receiving it for >120 days was 54% for individuals ages 70 to 74; 49% for individuals ages 75 to 79; and 46% for individuals ages 80 to 84.

This 3-part review summarizes findings and risk–benefit considerations when prescribing antipsychotics to older individuals. Part 1 focused on those with chronic psychotic disorders, such as schizophrenia or bipolar disorder,³ and part 3 will cover patients with dementia. This review (part 2) aims to:

• briefly summarize the results of randomized controlled trials (RCTs) of second-generation antipsychotics (SGAs) and other major studies and analyses in older patients with major depressive disorder (MDD)
• provide a summative opinion on the relative risks and benefits associated with using antipsychotics in older adults with MDD
• highlight the gaps in the evidence base and areas that need additional research.

Summary of benefits, place in treatment armamentarium

The prevalence of MDD and clinically significant depressive symptoms in community­dwelling older adults is 3% to 4% and 15%, respectively, and as high as 16% and 50%, respectively, in nursing home residents.⁴ Because late-life depression is associated with suffering, disability, and excessive mortality, it needs to be recognized and treated aggressively.⁵ Antidepressants are the mainstay of pharmacotherapy for late-life depression. Guidelines and expert opinion informed by the current evidence recommend using selective serotonin reuptake inhibitors, such as escitalopram or sertraline, as a first-line treatment; serotonin norepinephrine reuptake inhibitors, such as duloxetine or venlafaxine, as a second-line treatment; and other antidepressants, such as bupropion or nortriptyline, as a third-line treatment.^5,6 However, antipsychotics also have a role in treating late-life depression.

Over the past decade, several anti­psychotics have been FDA-approved for treating MDD: aripiprazole and brexpiprazole as adjunctive treatment of MDD in adults; olanzapine-fluoxetine combination for acute and maintenance treatment of treatment-resistant depression in adults and geriatric adults; and quetiapine extended-release (XR) as monotherapy for MDD in adults and as adjunctive treatment of MDD in adults and geriatric adults who have had an inadequate response to antidepressants alone (Table 1). However, “black-box” warnings for all first-generation antipsychotics (FGAs) and SGAs alert clinicians that these medications have been associated with serious adverse events in older adults with dementia, including “deaths […] due to heart-related events (eg, heart failure, sudden death) or infections (mostly pneumonia),” with 15 of 17 placebo-controlled trials showing a higher number of deaths with an antipsychotic compared with placebo.⁷ Although similar controlled data on the mortality risk of antipsychotics in older adults with mood disorders do not exist, most experts limit their use to 2 groups of older patients: those with MDD and psychotic features (“psychotic depression”) and those with treatment-resistant depression.

Clinical trials

Olanzapine plus sertraline as first-line pharmacotherapy for MDD with psychotic features. Meyers et al¹¹ reported on a double-blind randomized comparison of olanzapine plus placebo vs olanzapine plus sertraline in 259 patients with MDD with psychotic features. An unusual feature of this trial is that it included a similar number of younger and older participants (ages 18 to 93): 117 participants were age <60 (mean age [standard deviation (SD)]: 41.3 [10.8]) and 142 were age ≥60 (mean age [SD]: 71.7 [7.8]). The same dose titration schedules based on efficacy and tolerability were used in both younger and older participants. At the end of the study, the mean dose (SD) of sertraline (or placebo) did not differ significantly in younger (174.3 mg/d [34.1]) and older participants (165.7 mg/d [43.4]). However, the mean dose (SD) of olanzapine was significantly higher in younger patients (15.7 mg/d [4.7]) than in older participants (13.4 mg/d [5.1]).

In both age groups, olanzapine plus sertraline was more efficacious than olanzapine plus placebo, and there was no statistical interaction between age, time, and treatment group (ie, the trajectories of improvement were similar in older and younger patients receiving either olanzapine or olanzapine plus sertraline). Similarly, drop-out rates because of poor tolerability did not differ significantly in younger (4.3%) and older participants (5.6%). However, in a multinomial regression, older participants were more likely to discontinue treatment because of poor tolerability.²² Older participants were significantly less likely to experience weight gain (mean [SD]: +3.3 [4.9] vs +6.5 [6.6] kg) or an increase in fasting glucose and more likely to experience a fall, pedal edema, or extrapyramidal symptoms.^11,22-24 Cholesterol and triglyceride increased significantly and similarly in both age groups. The incidence of symptoms of tardive dyskinesia (TD) over the 12-week trial was low (<5%) in both younger and older participants, and clinically diagnosed TD was reported in only 1 (older) participant.²⁵

Venlafaxine plus aripiprazole for treatment-resistant MDD. In the largest double-blind randomized study of augmentation pharmacotherapy for late-life treatment-resistant depression published to date, Lenze et al²¹ compared venlafaxine plus aripiprazole vs venlafaxine plus placebo in 181 patients age >60 (mean age 66, with 49 participants age >70) with MDD who did not remit after 12 weeks of treatment with venlafaxine (up to 300 mg/d). After 12 weeks of augmentation, remission rates were significantly higher with aripiprazole than with placebo: 40 (44%) vs 26 (29%); odds ratio (95% confidence interval [CI]): 2.0 (1.1 to 3.7). The median final aripiprazole dose was 7 mg/d (range 2 to 15 mg/d) in remitters and 10 mg/d (range 2 to 15 mg/d) in nonremitters.

Five of 90 participants (5%) discontinued aripiprazole (1 each: suicide, jitteriness/akathisia, worsening parkinsonism; and 2 withdrew consent); 8 of 90 (9%) discontinued placebo (2 each: lack of efficacy, headache; 1: worsening parkinsonism; and 3 withdrew consent). The completed suicide occurred after 5 weeks of treatment with aripiprazole and was judged to be “neither due to emergent suicidal ideation nor to aripiprazole side-effects, but was concluded by investigators to be a result of the individual’s persisting and long-standing suicidal ideation.”²¹ Including the suicide, there were 4 serious adverse events (5%) in those receiving aripiprazole (1 each: suicide, congestive heart failure, mild stroke, and diverticulitis) and 2 (2%) in those receiving placebo (1 each: myocardial infarction, hospitalized for vomiting due to accidentally taking extra venlafaxine). In 86 participants receiving aripiprazole and 87 receiving placebo, the most frequently reported adverse effects were increased dream activity (aripiprazole: 23 [27%] vs placebo: 12 [14%]), weight gain (17 [20%] vs 8 [9%]), and tremor (5 [6%] vs 0). Akathisia and parkinsonism were observed more frequently with aripiprazole than with placebo (akathisia: 24 [26%] of 91 vs 11 [12%] of 90; parkinsonism: 15 [17%] of 86 vs 2 [2%] of 81). Akathisia was generally mild and resolved with dose adjustment; however, it was associated with a transient increase in suicidality in 3 (3%) participants receiving aripiprazole vs 0 receiving placebo and persisted at the end of the trial in 5 (5%) participants receiving aripiprazole vs 2 (2%) receiving placebo. Participants receiving aripiprazole had a significantly larger increase in weight (mean [SD]: +1.93 [3.00] vs +0.01 [3.15] kg), but there were no differences between aripiprazole and placebo in changes in body fat, total cholesterol, high-density lipoprotein, low-density lipoprotein, triglycerides, glucose, insulin concentration, or QTc.

Citalopram plus risperidone for treatment-resistant MDD. Alexopoulos et al²⁶ reported an analysis of data from 110 patients age ≥55 years (mean age [SD]: 63.4 [4.8]), among 489 mixed-age patients with MDD. Participants (n = 110) who did not respond to 1 to 3 antidepressants (venlafaxine, sertraline, mirtazapine, fluoxetine, paroxetine, or bupropion in >90%) during their current depressive episode completed 4 to 6 weeks of treatment with citalopram up to 40 mg/d; 93 did not respond and were treated with open-label risperidone (0.25 to 1 mg/d) augmentation for 4 to 6 weeks. Sixty-three (68%) of these 93 patients remitted and were randomized to 24 weeks of double-blind continuation treatment with citalopram plus risperidone vs citalopram plus placebo. Neither the median times to relapse (105 vs 57 days) nor the relapse rates (risperidone: 18 of 32 [56%] vs placebo: 20 of 31 [65%]) differed significantly. During the open-label risperidone augmentation, the most common adverse events were dizziness and dry mouth (n = 9 each, 9.7% of 93). During the continuation phase, headache (n = 3; 9.1% of 32) was observed with risperidone but not with placebo (n = 0). There was no incident parkinsonism or abnormal movements noted, but risperidone was associated with weight gain during both the open-label risperidone augmentation phase (mean [SD]: +0.9 [2.1] kg) and the continuation phase (risperidone: +0.8 [3.5] vs placebo: −0.3 [2.8] kg).

Quetiapine XR monotherapy for MDD. Katila et al²⁷ reported on a placebo-controlled RCT of quetiapine XR (median dose, 158.7 mg/d; range, 50 to 300 mg/d) in 338 patients age ≥66 years (mean age [SD], 71.3 [7.5]) presenting with MDD and a major depressive episode with a duration <1 year and no history of failed antidepressants trials from 2 classes (more than two-thirds of participants had not received treatment). After 9 weeks, the reduction in depressive symptoms on the Montgomery-Åsberg Depression Rating Scale was significantly larger with quetiapine XR than with placebo (mean [SD]: −16.0 [9.3] vs −9.0 [9.9]). There were congruent, significant differences between quetiapine and placebo in terms of response rate (quetiapine XR: 105 of 164 [64%] vs placebo: 52 of 171 [30.4%]) and remission rate (92 of 164 [56.1%] vs placebo: 40 of 171 [23.4%]). The drop-out rates for all causes were similar, but the drop-out rate attributed to adverse events was higher with quetiapine than placebo (16 of 166 [9.6%] vs 7 of 172 [4.1%]). Most quetiapine drop-outs were attributable to dizziness, headache, and somnolence (n = 4 each), and placebo drop-outs were because of headache (n = 2). Consistent with the profile of quetiapine, adverse events with a rate that was at least 5% higher with quetiapine than with placebo included somnolence (64 of 166 [38.6%] vs 16 of 172 [9.3%]), dry mouth (34 [20.5%] vs 18 [10.5%]), and extrapyramidal symptoms (12 [7.2%] vs 4 [2.3%]). Changes in weight and laboratory test results (eg, glucose, lipid profile) were minimal and not clinically meaningful.

Other clinical data. The efficacy and relatively good tolerability of aripiprazole in older patients with treatment-resistant depression observed in the RCT by Lenze et al²¹ is congruent with the earlier results of 2 small (N = 20 and 24) pilot studies.^18,19 In both studies, the remission rate was 50%, and the most prevalent adverse effects were agitation/restlessness/akathisia or drowsiness/sedation. Similarly, in a post hoc pooled analysis of 409 participants ages 50 to 67 from 3 placebo-controlled randomized trials, the remission rate was significantly higher with aripiprazole than with placebo (32.5% vs 17.1%), and the most common adverse effects were akathisia or restlessness (64 of 210 [30.4%]), somnolence (18 [8.6%]), and insomnia (17 [8.1%]).²⁰

Clinical considerations

When assessing the relative benefits and risks of antipsychotics in older patients, it is important to remember that conclusions and summative opinions are necessarily influenced by the source of the data. Because much of what we know about the use of antipsychotics in geriatric adults is from clinical trials, we know more about their acute efficacy and tolerability than their long-term effectiveness and safety.²⁸ There are similar issues regarding the role of antipsychotics in treating MDD in late life. Based on the results of several RCTs,^8,11 a combination of an antidepressant plus an antipsychotic is the recommended pharmacotherapy for the acute treatment of MDD with psychotic features (Table 2).^{8,11,12,19-21,23-27} However, there are no published data to guide how long the antipsychotic should be continued.²⁹

In older patients with MDD without psychotic features, 1 relatively large placebo-controlled RCT,²¹ 2 smaller open studies,^18,19 and a post hoc analysis of a large placebo-controlled RCT in mixed-age adults²⁰ support the efficacy and relatively good tolerability of aripiprazole augmentation of an antidepressant for treatment-resistant MDD. Similarly, 1 large placebo-controlled RCT supports the efficacy and relatively good tolerability of quetiapine for non–treatment-resistant MDD. However, there are no comparative data assessing the relative merits of using these antipsy­chotics vs other pharmacologic strategies (eg, switching to another antidepressant, lithium augmentation, or combination of 2 antidepressants). Because older patients are more likely to experience adverse effects that may have more serious consequences (Table 3), many prudent clinicians reserve using antipsychotics as a third-line treatment in older patients with MDD without psychotic features and limit the duration of their use to a few months.³⁰

The proportion of older adults in the world population is growing rapidly. In the next 10 to 15 years, the population age >60 will grow 3.5 times more rapidly than the general population.¹ As a result, there is an increased urgency in examining benefits vs risks of antipsychotics in older individuals. In a 2010 U.S. nationally representative observational study, antipsychotic use was observed to rise slowly during early and middle adulthood, peaking at approximately age 55, declining slightly between ages 55 and 65, and then rising again after age 65, with >2% of individuals ages 80 to 84 receiving an anti­psychotic.² This is likely due to the chronology of psychotic, mood, and neurocognitive disorders across the life span. In this large national study, long-term antipsychotic treatment was common, and older patients were more likely to receive their prescriptions from non-psychiatrist physicians than from psychiatrists.² Among patients receiving an antipsychotic, the proportion of those receiving it for >120 days was 54% for individuals ages 70 to 74; 49% for individuals ages 75 to 79; and 46% for individuals ages 80 to 84.

This 3-part review summarizes findings and risk–benefit considerations when prescribing antipsychotics to older individuals. Part 1 focused on those with chronic psychotic disorders, such as schizophrenia or bipolar disorder,³ and part 3 will cover patients with dementia. This review (part 2) aims to:

• briefly summarize the results of randomized controlled trials (RCTs) of second-generation antipsychotics (SGAs) and other major studies and analyses in older patients with major depressive disorder (MDD)
• provide a summative opinion on the relative risks and benefits associated with using antipsychotics in older adults with MDD
• highlight the gaps in the evidence base and areas that need additional research.

Summary of benefits, place in treatment armamentarium

The prevalence of MDD and clinically significant depressive symptoms in community­dwelling older adults is 3% to 4% and 15%, respectively, and as high as 16% and 50%, respectively, in nursing home residents.⁴ Because late-life depression is associated with suffering, disability, and excessive mortality, it needs to be recognized and treated aggressively.⁵ Antidepressants are the mainstay of pharmacotherapy for late-life depression. Guidelines and expert opinion informed by the current evidence recommend using selective serotonin reuptake inhibitors, such as escitalopram or sertraline, as a first-line treatment; serotonin norepinephrine reuptake inhibitors, such as duloxetine or venlafaxine, as a second-line treatment; and other antidepressants, such as bupropion or nortriptyline, as a third-line treatment.^5,6 However, antipsychotics also have a role in treating late-life depression.

Over the past decade, several anti­psychotics have been FDA-approved for treating MDD: aripiprazole and brexpiprazole as adjunctive treatment of MDD in adults; olanzapine-fluoxetine combination for acute and maintenance treatment of treatment-resistant depression in adults and geriatric adults; and quetiapine extended-release (XR) as monotherapy for MDD in adults and as adjunctive treatment of MDD in adults and geriatric adults who have had an inadequate response to antidepressants alone (Table 1). However, “black-box” warnings for all first-generation antipsychotics (FGAs) and SGAs alert clinicians that these medications have been associated with serious adverse events in older adults with dementia, including “deaths […] due to heart-related events (eg, heart failure, sudden death) or infections (mostly pneumonia),” with 15 of 17 placebo-controlled trials showing a higher number of deaths with an antipsychotic compared with placebo.⁷ Although similar controlled data on the mortality risk of antipsychotics in older adults with mood disorders do not exist, most experts limit their use to 2 groups of older patients: those with MDD and psychotic features (“psychotic depression”) and those with treatment-resistant depression.

Clinical trials

Olanzapine plus sertraline as first-line pharmacotherapy for MDD with psychotic features. Meyers et al¹¹ reported on a double-blind randomized comparison of olanzapine plus placebo vs olanzapine plus sertraline in 259 patients with MDD with psychotic features. An unusual feature of this trial is that it included a similar number of younger and older participants (ages 18 to 93): 117 participants were age <60 (mean age [standard deviation (SD)]: 41.3 [10.8]) and 142 were age ≥60 (mean age [SD]: 71.7 [7.8]). The same dose titration schedules based on efficacy and tolerability were used in both younger and older participants. At the end of the study, the mean dose (SD) of sertraline (or placebo) did not differ significantly in younger (174.3 mg/d [34.1]) and older participants (165.7 mg/d [43.4]). However, the mean dose (SD) of olanzapine was significantly higher in younger patients (15.7 mg/d [4.7]) than in older participants (13.4 mg/d [5.1]).

In both age groups, olanzapine plus sertraline was more efficacious than olanzapine plus placebo, and there was no statistical interaction between age, time, and treatment group (ie, the trajectories of improvement were similar in older and younger patients receiving either olanzapine or olanzapine plus sertraline). Similarly, drop-out rates because of poor tolerability did not differ significantly in younger (4.3%) and older participants (5.6%). However, in a multinomial regression, older participants were more likely to discontinue treatment because of poor tolerability.²² Older participants were significantly less likely to experience weight gain (mean [SD]: +3.3 [4.9] vs +6.5 [6.6] kg) or an increase in fasting glucose and more likely to experience a fall, pedal edema, or extrapyramidal symptoms.^11,22-24 Cholesterol and triglyceride increased significantly and similarly in both age groups. The incidence of symptoms of tardive dyskinesia (TD) over the 12-week trial was low (<5%) in both younger and older participants, and clinically diagnosed TD was reported in only 1 (older) participant.²⁵

Venlafaxine plus aripiprazole for treatment-resistant MDD. In the largest double-blind randomized study of augmentation pharmacotherapy for late-life treatment-resistant depression published to date, Lenze et al²¹ compared venlafaxine plus aripiprazole vs venlafaxine plus placebo in 181 patients age >60 (mean age 66, with 49 participants age >70) with MDD who did not remit after 12 weeks of treatment with venlafaxine (up to 300 mg/d). After 12 weeks of augmentation, remission rates were significantly higher with aripiprazole than with placebo: 40 (44%) vs 26 (29%); odds ratio (95% confidence interval [CI]): 2.0 (1.1 to 3.7). The median final aripiprazole dose was 7 mg/d (range 2 to 15 mg/d) in remitters and 10 mg/d (range 2 to 15 mg/d) in nonremitters.

Five of 90 participants (5%) discontinued aripiprazole (1 each: suicide, jitteriness/akathisia, worsening parkinsonism; and 2 withdrew consent); 8 of 90 (9%) discontinued placebo (2 each: lack of efficacy, headache; 1: worsening parkinsonism; and 3 withdrew consent). The completed suicide occurred after 5 weeks of treatment with aripiprazole and was judged to be “neither due to emergent suicidal ideation nor to aripiprazole side-effects, but was concluded by investigators to be a result of the individual’s persisting and long-standing suicidal ideation.”²¹ Including the suicide, there were 4 serious adverse events (5%) in those receiving aripiprazole (1 each: suicide, congestive heart failure, mild stroke, and diverticulitis) and 2 (2%) in those receiving placebo (1 each: myocardial infarction, hospitalized for vomiting due to accidentally taking extra venlafaxine). In 86 participants receiving aripiprazole and 87 receiving placebo, the most frequently reported adverse effects were increased dream activity (aripiprazole: 23 [27%] vs placebo: 12 [14%]), weight gain (17 [20%] vs 8 [9%]), and tremor (5 [6%] vs 0). Akathisia and parkinsonism were observed more frequently with aripiprazole than with placebo (akathisia: 24 [26%] of 91 vs 11 [12%] of 90; parkinsonism: 15 [17%] of 86 vs 2 [2%] of 81). Akathisia was generally mild and resolved with dose adjustment; however, it was associated with a transient increase in suicidality in 3 (3%) participants receiving aripiprazole vs 0 receiving placebo and persisted at the end of the trial in 5 (5%) participants receiving aripiprazole vs 2 (2%) receiving placebo. Participants receiving aripiprazole had a significantly larger increase in weight (mean [SD]: +1.93 [3.00] vs +0.01 [3.15] kg), but there were no differences between aripiprazole and placebo in changes in body fat, total cholesterol, high-density lipoprotein, low-density lipoprotein, triglycerides, glucose, insulin concentration, or QTc.

Citalopram plus risperidone for treatment-resistant MDD. Alexopoulos et al²⁶ reported an analysis of data from 110 patients age ≥55 years (mean age [SD]: 63.4 [4.8]), among 489 mixed-age patients with MDD. Participants (n = 110) who did not respond to 1 to 3 antidepressants (venlafaxine, sertraline, mirtazapine, fluoxetine, paroxetine, or bupropion in >90%) during their current depressive episode completed 4 to 6 weeks of treatment with citalopram up to 40 mg/d; 93 did not respond and were treated with open-label risperidone (0.25 to 1 mg/d) augmentation for 4 to 6 weeks. Sixty-three (68%) of these 93 patients remitted and were randomized to 24 weeks of double-blind continuation treatment with citalopram plus risperidone vs citalopram plus placebo. Neither the median times to relapse (105 vs 57 days) nor the relapse rates (risperidone: 18 of 32 [56%] vs placebo: 20 of 31 [65%]) differed significantly. During the open-label risperidone augmentation, the most common adverse events were dizziness and dry mouth (n = 9 each, 9.7% of 93). During the continuation phase, headache (n = 3; 9.1% of 32) was observed with risperidone but not with placebo (n = 0). There was no incident parkinsonism or abnormal movements noted, but risperidone was associated with weight gain during both the open-label risperidone augmentation phase (mean [SD]: +0.9 [2.1] kg) and the continuation phase (risperidone: +0.8 [3.5] vs placebo: −0.3 [2.8] kg).

Quetiapine XR monotherapy for MDD. Katila et al²⁷ reported on a placebo-controlled RCT of quetiapine XR (median dose, 158.7 mg/d; range, 50 to 300 mg/d) in 338 patients age ≥66 years (mean age [SD], 71.3 [7.5]) presenting with MDD and a major depressive episode with a duration <1 year and no history of failed antidepressants trials from 2 classes (more than two-thirds of participants had not received treatment). After 9 weeks, the reduction in depressive symptoms on the Montgomery-Åsberg Depression Rating Scale was significantly larger with quetiapine XR than with placebo (mean [SD]: −16.0 [9.3] vs −9.0 [9.9]). There were congruent, significant differences between quetiapine and placebo in terms of response rate (quetiapine XR: 105 of 164 [64%] vs placebo: 52 of 171 [30.4%]) and remission rate (92 of 164 [56.1%] vs placebo: 40 of 171 [23.4%]). The drop-out rates for all causes were similar, but the drop-out rate attributed to adverse events was higher with quetiapine than placebo (16 of 166 [9.6%] vs 7 of 172 [4.1%]). Most quetiapine drop-outs were attributable to dizziness, headache, and somnolence (n = 4 each), and placebo drop-outs were because of headache (n = 2). Consistent with the profile of quetiapine, adverse events with a rate that was at least 5% higher with quetiapine than with placebo included somnolence (64 of 166 [38.6%] vs 16 of 172 [9.3%]), dry mouth (34 [20.5%] vs 18 [10.5%]), and extrapyramidal symptoms (12 [7.2%] vs 4 [2.3%]). Changes in weight and laboratory test results (eg, glucose, lipid profile) were minimal and not clinically meaningful.

Other clinical data. The efficacy and relatively good tolerability of aripiprazole in older patients with treatment-resistant depression observed in the RCT by Lenze et al²¹ is congruent with the earlier results of 2 small (N = 20 and 24) pilot studies.^18,19 In both studies, the remission rate was 50%, and the most prevalent adverse effects were agitation/restlessness/akathisia or drowsiness/sedation. Similarly, in a post hoc pooled analysis of 409 participants ages 50 to 67 from 3 placebo-controlled randomized trials, the remission rate was significantly higher with aripiprazole than with placebo (32.5% vs 17.1%), and the most common adverse effects were akathisia or restlessness (64 of 210 [30.4%]), somnolence (18 [8.6%]), and insomnia (17 [8.1%]).²⁰

Clinical considerations

When assessing the relative benefits and risks of antipsychotics in older patients, it is important to remember that conclusions and summative opinions are necessarily influenced by the source of the data. Because much of what we know about the use of antipsychotics in geriatric adults is from clinical trials, we know more about their acute efficacy and tolerability than their long-term effectiveness and safety.²⁸ There are similar issues regarding the role of antipsychotics in treating MDD in late life. Based on the results of several RCTs,^8,11 a combination of an antidepressant plus an antipsychotic is the recommended pharmacotherapy for the acute treatment of MDD with psychotic features (Table 2).^{8,11,12,19-21,23-27} However, there are no published data to guide how long the antipsychotic should be continued.²⁹

In older patients with MDD without psychotic features, 1 relatively large placebo-controlled RCT,²¹ 2 smaller open studies,^18,19 and a post hoc analysis of a large placebo-controlled RCT in mixed-age adults²⁰ support the efficacy and relatively good tolerability of aripiprazole augmentation of an antidepressant for treatment-resistant MDD. Similarly, 1 large placebo-controlled RCT supports the efficacy and relatively good tolerability of quetiapine for non–treatment-resistant MDD. However, there are no comparative data assessing the relative merits of using these antipsy­chotics vs other pharmacologic strategies (eg, switching to another antidepressant, lithium augmentation, or combination of 2 antidepressants). Because older patients are more likely to experience adverse effects that may have more serious consequences (Table 3), many prudent clinicians reserve using antipsychotics as a third-line treatment in older patients with MDD without psychotic features and limit the duration of their use to a few months.³⁰

Antipsychotics are FDA-approved as a primary treatment for schizophrenia and bipolar disorder and as adjunctive therapy for major depressive disorder. In the United States, approximately 26% of antipsychotic prescriptions written for these indications are for individuals age >65.¹ Additionally, antipsychotics are widely used to treat behavioral symptoms associated with dementia.¹ The rapid expansion of the use of second-generation antipsychotics (SGAs), in particular, has been driven in part by their lower risk for extrapyramidal symptoms (EPS) compared with first-generation antipsychotics (FGAs).¹ However, a growing body of data indicates that all antipsychotics have a range of adverse effects in older patients. This focus is critical in light of demographic trends—in the next 10 to 15 years, the population age >60 will grow 3.5 times more rapidly than the general population.²

In this context, psychiatrists need information on the relative risks of antipsychotics for older patients. This 3-part series summarizes findings and recommendations on safety and tolerability when prescribing antipsychotics in older individuals with chronic psychotic disorders, such as schizophrenia, bipolar disorder, depression, and dementia. This review aims to:

• briefly summarize the major studies and analyses relevant to older patients with these diagnoses
• provide a summative opinion on safety and tolerability issues in these older adults
• highlight the gaps in the evidence base and areas that need additional research.

Part 1 focuses on older adults with schizophrenia or bipolar disorder. Subsequent articles will focus on prescribing antipsychotics to older adults with depression and those with dementia.

Schizophrenia

Summary of benefits, place in treatment armamentarium. Individuals with schizophrenia have a shorter life expectancy than that of the general population mostly as a result of suicide and comorbid physical illnesses,³ but the number of patients with schizophrenia age >55 will double over the next 2 decades.⁴ With aging, both positive and negative symptoms may be a focus of treatment (Table 1).^5,6 Antipsychotics are a first-line treatment for older patients with schizophrenia with few medication alternatives.⁷ Safety risks associated with antipsychotics in older people span a broad spectrum (Table 2).⁸

A 6-week prospective RCT evaluated paliperidone extended-release vs placebo in 114 older adults (age ≥65 years; mean age, 70 years) with schizophrenia.¹⁴ There was an optional 24-week extension of open-label treatment with paliperidone. Mean daily dose of paliperidone was 8.4 mg. Efficacy measures did not show consistent statistically significant differences between treatment groups. Discontinuation rates were similar between paliperidone (7%) vs placebo (8%). Serious adverse events occurred in 3% of paliperidone-treated vs 8% of placebo-treated patients. Elevated prolactin levels occurred in one-half of paliperidone-treated patients. There were no prolactin or glucose treatment-related adverse events or significant mean changes in body weight for either paliperidone-treated or placebo-treated patients. Safety findings in the 24-week, open-label extension group were consistent with the RCT results.

Howanitz et al¹⁵ conducted a 12-week, prospective RCT that compared clozapine (mean dose, 300 mg/d) with chlorpromazine (mean dose, 600 mg/d) in 42 older adults (mean age, 67 years) with schizophrenia. Drop-out rate prior to 5 weeks was 19% and similar between groups. Common adverse effects included sialorrhea, hematologic abnormalities, sedation, tachycardia, EPS, and weight gain. Although both drugs were effective, more patients taking clozapine had tachycardia and weight gain, while more chlorpromazine patients reported sedation.

There have been other, less rigorous studies.^7,8 Most of these studies evaluated risperidone and olanzapine, and most were conducted in “younger” geriatric patients (age <75 years). Although patients who participate in clinical trials may be healthier than “typical” patients, adverse effects such as EPS, sedation, and weight gain were still relatively common in these studies.

Other clinical data. A major consideration in treating older adults with schizophrenia is balancing the need to administer an antipsychotic dose high enough to alleviate psychotic symptoms while minimizing dose-dependent adverse effects. There is a U-shaped relationship between age and vulnerability to antipsychotic adverse effects,^16,17 wherein adverse effects are highest at younger and older ages. Evidence supports using the lowest effective antipsychotic dose for geriatric patients with schizophrenia. Positive emission tomography (PET) studies suggest that older patients develop EPS with lower doses despite lower receptor occupancy.^17,18 A recent study of 35 older patients (mean age, 60.1 years) with schizophrenia obtained PET, clinical measures, and blood pharmacokinetic measures before and after reduction of risperidone or olanzapine doses.¹⁸ A ≥40% reduction in dose was associated with reduced adverse effects, particularly EPS and elevation of prolactin levels. Moreover, the therapeutic window of striatal D2/D3 receptor occupancy appeared to be 50% to 60% in these older patients, compared with 65% to 80% in younger patients.

Long-term risks of antipsychotic treatment across the lifespan are less clear, with evidence suggesting both lower and higher mortality risk.^19,20 It is difficult to fully disentangle the long-term risks of antipsychotics from the cumulative effects of lifestyle and comorbidity among individuals who have lived with schizophrenia for decades. Large naturalistic studies that include substantial numbers of older people with schizophrenia might be a way to elicit more information on long-term safety. The Schizophrenia Outpatient Health Outcome (SOHO) study was a large naturalistic trial that recruited >10,000 individuals with schizophrenia in 10 European countries.²¹ Although the SOHO study found differences between antipsychotics and adverse effects, such as EPS, weight gain, and sexual dysfunction, because the mean age of these patients was approximately 40 years and the follow-up period was only 3 years, it is difficult to draw conclusions that could be relevant to older individuals who have had schizophrenia for decades.

Bipolar Disorder

A secondary analysis of mixed-age, RCTs examined response in older adults (age ≥55 years) with bipolar I depression who received lurasidone as monotherapy or adjunctive therapy.³² In the monotherapy study, these patients were randomized to 6 weeks of lurasidone 20 to 60 mg/d, lurasidone 80 to 120 mg/d, or placebo. In the adjunctive therapy study, they were randomized to lurasidone 20 to 120 mg/d or placebo with either lithium or valproate. There were 83 older adults (17.1% of the sample) in the monotherapy study and 53 (15.6%) in the adjunctive therapy study. Mean improvement in depression was significantly higher for both doses of lurasidone monotherapy than placebo. Adjunctive lurasidone was not associated with statistically significant improvement vs placebo. The most frequent adverse events in older patients on lurasidone monotherapy 20 to 60 mg/d or 80 to 120 mg/d were nausea (18.5% and 9.7%, respectively) and somnolence (11.1% and 0%, respectively). Akathisia (9.7%) and insomnia (9.7%) were the most common adverse events in the group receiving 80 to 120 mg/d, with the rate of akathisia exhibiting a dose-related increase. Weight change with lurasidone was similar to placebo, and there were no clinically meaningful group changes in vital signs, electrocardiography, or laboratory parameters.

A small (N = 20) open study found improvement in older adults with bipolar depression with aripiprazole (mean dose, 10.3 mg/d).³³ Adverse effects included restlessness and weight gain (n = 3, 9% each), sedation (n = 2, 10%), and drooling and diarrhea/loose stools (n = 1, 5% each). In another small study (N = 15) using asenapine (mean dose, 11.2 mg/d) in mainly older bipolar patients with depression, the most common adverse effects were gastrointestinal (GI) discomfort (n = 5, 33%) and restlessness, tremors, cognitive difficulties, and sluggishness (n = 2, 13% each).³⁴

Clinical trials: Bipolar mania. Researchers conducted a pooled analysis of two 12-week randomized trials comparing quetiapine with placebo in a mixed-age sample with bipolar mania.³⁵ In a subgroup of 59 older patients (mean age, 62.9 years), manic symptoms improved significantly more with quetiapine (modal dose, 550 mg/d) than with placebo. Adverse effects reported by >10% of older patients were dry mouth, somnolence, postural hypotension, insomnia, weight gain, and dizziness. Insomnia was reported by >10% of patients receiving placebo.

In a case series of 11 elderly patients with mania receiving asenapine, Baruch et al³⁶ reported a 63% remission rate. One patient discontinued the study because of a new rash, 1 discontinued after developing peripheral edema, and 3 patients reported mild sedation.

Beyer et al³⁷ reported on a post hoc analysis of 94 older adults (mean age, 57.1 years; range, 50.1 to 74.8 years) with acute bipolar mania receiving olanzapine (n = 47), divalproex (n = 31), or placebo (n = 16) in a pooled olanzapine clinical trials database. Patients receiving olanzapine or divalproex had improvement in mania; those receiving placebo did not improve. Safety findings were comparable with reports in younger patients with mania.

Other clinical data. Adverse effects found in mixed-age samples using secondary analyses of clinical trials need to be interpreted with caution because these types of studies usually exclude individuals with significant medical comorbidity. Medical burden, cognitive impairment, or concomitant medications generally necessitate slower drug titration and lower total daily dosing. For example, a secondary analysis of the U.S. National Institute of Health-funded Systematic Treatment Enhancement Program for Bipolar Disorder study, which had broader inclusion criteria than most clinical trials, reported that, although recovery rates in older adults with bipolar disorder were fairly good (78.5%), lower doses of risperidone were used in older vs younger patients.³⁸

Clinical considerations

Interpretation of the relative risks of antipsychotics in older people must be tempered by the caveat that there is limited high-quality data (Table 4). Antipsychotics are the first-line therapy for older patients with schizophrenia, although their use is supported by a small number of prospective RCTs. SGAs are preferred because of their lower propensity to cause EPS and other motor adverse effects. Older persons with schizophrenia have an EPS threshold lower than younger patients and determining the lowest effective dosage may minimize EPS and cognitive adverse effects. As individuals with long-standing schizophrenia get older, their antipsychotic dosages may need to be reduced, and clinicians need to monitor for adverse effects that are more common among older people, such as tardive dyskinesia and metabolic abnormalities. In healthy, “younger” geriatric patients, monitoring for adverse effects may be similar to monitoring of younger patients. Patients who are older or frail may need more frequent assessment.

Like older adults with schizophrenia, geriatric patients with bipolar disorder have reduced drug tolerability and experience more adverse effects than younger patients. There are no prospective controlled studies that evaluated using antipsychotics in older patients with bipolar disorder. In older bipolar patients, the most problematic adverse effects of antipsychotics are akathisia, parkinsonism, other EPS, sedation and dizziness (which may increase fall risk), and GI discomfort. A key tolerability and safety consideration when treating older adults with bipolar disorder is the role of antipsychotics in relation to the use of lithium and mood stabilizers. Some studies have suggested that lithium has neuroprotective effects when used long-term; however, at least 1 report suggested that long-term antipsychotic treatment may be associated with neurodegeneration.³⁹

The literature does not provide strong evidence on the many clinical variations that we see in routine practice settings, such as combinations of drug treatments or drugs prescribed to patients with specific comorbid conditions. There is a need for large cohort studies that monitor treatment course, medical comorbidity, and prognosis. Additionally, well-designed clinical trials such as the DART-AD, which investigated longer-term trajectories of people with dementia taking antipsychotics, should serve as a model for the type of research that is needed to better understand outcome variability among older people with chronic psychotic or bipolar disorders.⁴⁰

Antipsychotics are FDA-approved as a primary treatment for schizophrenia and bipolar disorder and as adjunctive therapy for major depressive disorder. In the United States, approximately 26% of antipsychotic prescriptions written for these indications are for individuals age >65.¹ Additionally, antipsychotics are widely used to treat behavioral symptoms associated with dementia.¹ The rapid expansion of the use of second-generation antipsychotics (SGAs), in particular, has been driven in part by their lower risk for extrapyramidal symptoms (EPS) compared with first-generation antipsychotics (FGAs).¹ However, a growing body of data indicates that all antipsychotics have a range of adverse effects in older patients. This focus is critical in light of demographic trends—in the next 10 to 15 years, the population age >60 will grow 3.5 times more rapidly than the general population.²

In this context, psychiatrists need information on the relative risks of antipsychotics for older patients. This 3-part series summarizes findings and recommendations on safety and tolerability when prescribing antipsychotics in older individuals with chronic psychotic disorders, such as schizophrenia, bipolar disorder, depression, and dementia. This review aims to:

• briefly summarize the major studies and analyses relevant to older patients with these diagnoses
• provide a summative opinion on safety and tolerability issues in these older adults
• highlight the gaps in the evidence base and areas that need additional research.

Part 1 focuses on older adults with schizophrenia or bipolar disorder. Subsequent articles will focus on prescribing antipsychotics to older adults with depression and those with dementia.

Schizophrenia

Summary of benefits, place in treatment armamentarium. Individuals with schizophrenia have a shorter life expectancy than that of the general population mostly as a result of suicide and comorbid physical illnesses,³ but the number of patients with schizophrenia age >55 will double over the next 2 decades.⁴ With aging, both positive and negative symptoms may be a focus of treatment (Table 1).^5,6 Antipsychotics are a first-line treatment for older patients with schizophrenia with few medication alternatives.⁷ Safety risks associated with antipsychotics in older people span a broad spectrum (Table 2).⁸

A 6-week prospective RCT evaluated paliperidone extended-release vs placebo in 114 older adults (age ≥65 years; mean age, 70 years) with schizophrenia.¹⁴ There was an optional 24-week extension of open-label treatment with paliperidone. Mean daily dose of paliperidone was 8.4 mg. Efficacy measures did not show consistent statistically significant differences between treatment groups. Discontinuation rates were similar between paliperidone (7%) vs placebo (8%). Serious adverse events occurred in 3% of paliperidone-treated vs 8% of placebo-treated patients. Elevated prolactin levels occurred in one-half of paliperidone-treated patients. There were no prolactin or glucose treatment-related adverse events or significant mean changes in body weight for either paliperidone-treated or placebo-treated patients. Safety findings in the 24-week, open-label extension group were consistent with the RCT results.

Howanitz et al¹⁵ conducted a 12-week, prospective RCT that compared clozapine (mean dose, 300 mg/d) with chlorpromazine (mean dose, 600 mg/d) in 42 older adults (mean age, 67 years) with schizophrenia. Drop-out rate prior to 5 weeks was 19% and similar between groups. Common adverse effects included sialorrhea, hematologic abnormalities, sedation, tachycardia, EPS, and weight gain. Although both drugs were effective, more patients taking clozapine had tachycardia and weight gain, while more chlorpromazine patients reported sedation.

There have been other, less rigorous studies.^7,8 Most of these studies evaluated risperidone and olanzapine, and most were conducted in “younger” geriatric patients (age <75 years). Although patients who participate in clinical trials may be healthier than “typical” patients, adverse effects such as EPS, sedation, and weight gain were still relatively common in these studies.

Other clinical data. A major consideration in treating older adults with schizophrenia is balancing the need to administer an antipsychotic dose high enough to alleviate psychotic symptoms while minimizing dose-dependent adverse effects. There is a U-shaped relationship between age and vulnerability to antipsychotic adverse effects,^16,17 wherein adverse effects are highest at younger and older ages. Evidence supports using the lowest effective antipsychotic dose for geriatric patients with schizophrenia. Positive emission tomography (PET) studies suggest that older patients develop EPS with lower doses despite lower receptor occupancy.^17,18 A recent study of 35 older patients (mean age, 60.1 years) with schizophrenia obtained PET, clinical measures, and blood pharmacokinetic measures before and after reduction of risperidone or olanzapine doses.¹⁸ A ≥40% reduction in dose was associated with reduced adverse effects, particularly EPS and elevation of prolactin levels. Moreover, the therapeutic window of striatal D2/D3 receptor occupancy appeared to be 50% to 60% in these older patients, compared with 65% to 80% in younger patients.

Long-term risks of antipsychotic treatment across the lifespan are less clear, with evidence suggesting both lower and higher mortality risk.^19,20 It is difficult to fully disentangle the long-term risks of antipsychotics from the cumulative effects of lifestyle and comorbidity among individuals who have lived with schizophrenia for decades. Large naturalistic studies that include substantial numbers of older people with schizophrenia might be a way to elicit more information on long-term safety. The Schizophrenia Outpatient Health Outcome (SOHO) study was a large naturalistic trial that recruited >10,000 individuals with schizophrenia in 10 European countries.²¹ Although the SOHO study found differences between antipsychotics and adverse effects, such as EPS, weight gain, and sexual dysfunction, because the mean age of these patients was approximately 40 years and the follow-up period was only 3 years, it is difficult to draw conclusions that could be relevant to older individuals who have had schizophrenia for decades.

Bipolar Disorder

A secondary analysis of mixed-age, RCTs examined response in older adults (age ≥55 years) with bipolar I depression who received lurasidone as monotherapy or adjunctive therapy.³² In the monotherapy study, these patients were randomized to 6 weeks of lurasidone 20 to 60 mg/d, lurasidone 80 to 120 mg/d, or placebo. In the adjunctive therapy study, they were randomized to lurasidone 20 to 120 mg/d or placebo with either lithium or valproate. There were 83 older adults (17.1% of the sample) in the monotherapy study and 53 (15.6%) in the adjunctive therapy study. Mean improvement in depression was significantly higher for both doses of lurasidone monotherapy than placebo. Adjunctive lurasidone was not associated with statistically significant improvement vs placebo. The most frequent adverse events in older patients on lurasidone monotherapy 20 to 60 mg/d or 80 to 120 mg/d were nausea (18.5% and 9.7%, respectively) and somnolence (11.1% and 0%, respectively). Akathisia (9.7%) and insomnia (9.7%) were the most common adverse events in the group receiving 80 to 120 mg/d, with the rate of akathisia exhibiting a dose-related increase. Weight change with lurasidone was similar to placebo, and there were no clinically meaningful group changes in vital signs, electrocardiography, or laboratory parameters.

A small (N = 20) open study found improvement in older adults with bipolar depression with aripiprazole (mean dose, 10.3 mg/d).³³ Adverse effects included restlessness and weight gain (n = 3, 9% each), sedation (n = 2, 10%), and drooling and diarrhea/loose stools (n = 1, 5% each). In another small study (N = 15) using asenapine (mean dose, 11.2 mg/d) in mainly older bipolar patients with depression, the most common adverse effects were gastrointestinal (GI) discomfort (n = 5, 33%) and restlessness, tremors, cognitive difficulties, and sluggishness (n = 2, 13% each).³⁴

Clinical trials: Bipolar mania. Researchers conducted a pooled analysis of two 12-week randomized trials comparing quetiapine with placebo in a mixed-age sample with bipolar mania.³⁵ In a subgroup of 59 older patients (mean age, 62.9 years), manic symptoms improved significantly more with quetiapine (modal dose, 550 mg/d) than with placebo. Adverse effects reported by >10% of older patients were dry mouth, somnolence, postural hypotension, insomnia, weight gain, and dizziness. Insomnia was reported by >10% of patients receiving placebo.

In a case series of 11 elderly patients with mania receiving asenapine, Baruch et al³⁶ reported a 63% remission rate. One patient discontinued the study because of a new rash, 1 discontinued after developing peripheral edema, and 3 patients reported mild sedation.

Beyer et al³⁷ reported on a post hoc analysis of 94 older adults (mean age, 57.1 years; range, 50.1 to 74.8 years) with acute bipolar mania receiving olanzapine (n = 47), divalproex (n = 31), or placebo (n = 16) in a pooled olanzapine clinical trials database. Patients receiving olanzapine or divalproex had improvement in mania; those receiving placebo did not improve. Safety findings were comparable with reports in younger patients with mania.

Other clinical data. Adverse effects found in mixed-age samples using secondary analyses of clinical trials need to be interpreted with caution because these types of studies usually exclude individuals with significant medical comorbidity. Medical burden, cognitive impairment, or concomitant medications generally necessitate slower drug titration and lower total daily dosing. For example, a secondary analysis of the U.S. National Institute of Health-funded Systematic Treatment Enhancement Program for Bipolar Disorder study, which had broader inclusion criteria than most clinical trials, reported that, although recovery rates in older adults with bipolar disorder were fairly good (78.5%), lower doses of risperidone were used in older vs younger patients.³⁸

Clinical considerations

Interpretation of the relative risks of antipsychotics in older people must be tempered by the caveat that there is limited high-quality data (Table 4). Antipsychotics are the first-line therapy for older patients with schizophrenia, although their use is supported by a small number of prospective RCTs. SGAs are preferred because of their lower propensity to cause EPS and other motor adverse effects. Older persons with schizophrenia have an EPS threshold lower than younger patients and determining the lowest effective dosage may minimize EPS and cognitive adverse effects. As individuals with long-standing schizophrenia get older, their antipsychotic dosages may need to be reduced, and clinicians need to monitor for adverse effects that are more common among older people, such as tardive dyskinesia and metabolic abnormalities. In healthy, “younger” geriatric patients, monitoring for adverse effects may be similar to monitoring of younger patients. Patients who are older or frail may need more frequent assessment.

Like older adults with schizophrenia, geriatric patients with bipolar disorder have reduced drug tolerability and experience more adverse effects than younger patients. There are no prospective controlled studies that evaluated using antipsychotics in older patients with bipolar disorder. In older bipolar patients, the most problematic adverse effects of antipsychotics are akathisia, parkinsonism, other EPS, sedation and dizziness (which may increase fall risk), and GI discomfort. A key tolerability and safety consideration when treating older adults with bipolar disorder is the role of antipsychotics in relation to the use of lithium and mood stabilizers. Some studies have suggested that lithium has neuroprotective effects when used long-term; however, at least 1 report suggested that long-term antipsychotic treatment may be associated with neurodegeneration.³⁹

The literature does not provide strong evidence on the many clinical variations that we see in routine practice settings, such as combinations of drug treatments or drugs prescribed to patients with specific comorbid conditions. There is a need for large cohort studies that monitor treatment course, medical comorbidity, and prognosis. Additionally, well-designed clinical trials such as the DART-AD, which investigated longer-term trajectories of people with dementia taking antipsychotics, should serve as a model for the type of research that is needed to better understand outcome variability among older people with chronic psychotic or bipolar disorders.⁴⁰

Antipsychotics are FDA-approved as a primary treatment for schizophrenia and bipolar disorder and as adjunctive therapy for major depressive disorder. In the United States, approximately 26% of antipsychotic prescriptions written for these indications are for individuals age >65.¹ Additionally, antipsychotics are widely used to treat behavioral symptoms associated with dementia.¹ The rapid expansion of the use of second-generation antipsychotics (SGAs), in particular, has been driven in part by their lower risk for extrapyramidal symptoms (EPS) compared with first-generation antipsychotics (FGAs).¹ However, a growing body of data indicates that all antipsychotics have a range of adverse effects in older patients. This focus is critical in light of demographic trends—in the next 10 to 15 years, the population age >60 will grow 3.5 times more rapidly than the general population.²

In this context, psychiatrists need information on the relative risks of antipsychotics for older patients. This 3-part series summarizes findings and recommendations on safety and tolerability when prescribing antipsychotics in older individuals with chronic psychotic disorders, such as schizophrenia, bipolar disorder, depression, and dementia. This review aims to:

• briefly summarize the major studies and analyses relevant to older patients with these diagnoses
• provide a summative opinion on safety and tolerability issues in these older adults
• highlight the gaps in the evidence base and areas that need additional research.

Part 1 focuses on older adults with schizophrenia or bipolar disorder. Subsequent articles will focus on prescribing antipsychotics to older adults with depression and those with dementia.

Schizophrenia

Summary of benefits, place in treatment armamentarium. Individuals with schizophrenia have a shorter life expectancy than that of the general population mostly as a result of suicide and comorbid physical illnesses,³ but the number of patients with schizophrenia age >55 will double over the next 2 decades.⁴ With aging, both positive and negative symptoms may be a focus of treatment (Table 1).^5,6 Antipsychotics are a first-line treatment for older patients with schizophrenia with few medication alternatives.⁷ Safety risks associated with antipsychotics in older people span a broad spectrum (Table 2).⁸

A 6-week prospective RCT evaluated paliperidone extended-release vs placebo in 114 older adults (age ≥65 years; mean age, 70 years) with schizophrenia.¹⁴ There was an optional 24-week extension of open-label treatment with paliperidone. Mean daily dose of paliperidone was 8.4 mg. Efficacy measures did not show consistent statistically significant differences between treatment groups. Discontinuation rates were similar between paliperidone (7%) vs placebo (8%). Serious adverse events occurred in 3% of paliperidone-treated vs 8% of placebo-treated patients. Elevated prolactin levels occurred in one-half of paliperidone-treated patients. There were no prolactin or glucose treatment-related adverse events or significant mean changes in body weight for either paliperidone-treated or placebo-treated patients. Safety findings in the 24-week, open-label extension group were consistent with the RCT results.

Howanitz et al¹⁵ conducted a 12-week, prospective RCT that compared clozapine (mean dose, 300 mg/d) with chlorpromazine (mean dose, 600 mg/d) in 42 older adults (mean age, 67 years) with schizophrenia. Drop-out rate prior to 5 weeks was 19% and similar between groups. Common adverse effects included sialorrhea, hematologic abnormalities, sedation, tachycardia, EPS, and weight gain. Although both drugs were effective, more patients taking clozapine had tachycardia and weight gain, while more chlorpromazine patients reported sedation.

There have been other, less rigorous studies.^7,8 Most of these studies evaluated risperidone and olanzapine, and most were conducted in “younger” geriatric patients (age <75 years). Although patients who participate in clinical trials may be healthier than “typical” patients, adverse effects such as EPS, sedation, and weight gain were still relatively common in these studies.

Other clinical data. A major consideration in treating older adults with schizophrenia is balancing the need to administer an antipsychotic dose high enough to alleviate psychotic symptoms while minimizing dose-dependent adverse effects. There is a U-shaped relationship between age and vulnerability to antipsychotic adverse effects,^16,17 wherein adverse effects are highest at younger and older ages. Evidence supports using the lowest effective antipsychotic dose for geriatric patients with schizophrenia. Positive emission tomography (PET) studies suggest that older patients develop EPS with lower doses despite lower receptor occupancy.^17,18 A recent study of 35 older patients (mean age, 60.1 years) with schizophrenia obtained PET, clinical measures, and blood pharmacokinetic measures before and after reduction of risperidone or olanzapine doses.¹⁸ A ≥40% reduction in dose was associated with reduced adverse effects, particularly EPS and elevation of prolactin levels. Moreover, the therapeutic window of striatal D2/D3 receptor occupancy appeared to be 50% to 60% in these older patients, compared with 65% to 80% in younger patients.

Long-term risks of antipsychotic treatment across the lifespan are less clear, with evidence suggesting both lower and higher mortality risk.^19,20 It is difficult to fully disentangle the long-term risks of antipsychotics from the cumulative effects of lifestyle and comorbidity among individuals who have lived with schizophrenia for decades. Large naturalistic studies that include substantial numbers of older people with schizophrenia might be a way to elicit more information on long-term safety. The Schizophrenia Outpatient Health Outcome (SOHO) study was a large naturalistic trial that recruited >10,000 individuals with schizophrenia in 10 European countries.²¹ Although the SOHO study found differences between antipsychotics and adverse effects, such as EPS, weight gain, and sexual dysfunction, because the mean age of these patients was approximately 40 years and the follow-up period was only 3 years, it is difficult to draw conclusions that could be relevant to older individuals who have had schizophrenia for decades.

Bipolar Disorder

A secondary analysis of mixed-age, RCTs examined response in older adults (age ≥55 years) with bipolar I depression who received lurasidone as monotherapy or adjunctive therapy.³² In the monotherapy study, these patients were randomized to 6 weeks of lurasidone 20 to 60 mg/d, lurasidone 80 to 120 mg/d, or placebo. In the adjunctive therapy study, they were randomized to lurasidone 20 to 120 mg/d or placebo with either lithium or valproate. There were 83 older adults (17.1% of the sample) in the monotherapy study and 53 (15.6%) in the adjunctive therapy study. Mean improvement in depression was significantly higher for both doses of lurasidone monotherapy than placebo. Adjunctive lurasidone was not associated with statistically significant improvement vs placebo. The most frequent adverse events in older patients on lurasidone monotherapy 20 to 60 mg/d or 80 to 120 mg/d were nausea (18.5% and 9.7%, respectively) and somnolence (11.1% and 0%, respectively). Akathisia (9.7%) and insomnia (9.7%) were the most common adverse events in the group receiving 80 to 120 mg/d, with the rate of akathisia exhibiting a dose-related increase. Weight change with lurasidone was similar to placebo, and there were no clinically meaningful group changes in vital signs, electrocardiography, or laboratory parameters.

A small (N = 20) open study found improvement in older adults with bipolar depression with aripiprazole (mean dose, 10.3 mg/d).³³ Adverse effects included restlessness and weight gain (n = 3, 9% each), sedation (n = 2, 10%), and drooling and diarrhea/loose stools (n = 1, 5% each). In another small study (N = 15) using asenapine (mean dose, 11.2 mg/d) in mainly older bipolar patients with depression, the most common adverse effects were gastrointestinal (GI) discomfort (n = 5, 33%) and restlessness, tremors, cognitive difficulties, and sluggishness (n = 2, 13% each).³⁴

Clinical trials: Bipolar mania. Researchers conducted a pooled analysis of two 12-week randomized trials comparing quetiapine with placebo in a mixed-age sample with bipolar mania.³⁵ In a subgroup of 59 older patients (mean age, 62.9 years), manic symptoms improved significantly more with quetiapine (modal dose, 550 mg/d) than with placebo. Adverse effects reported by >10% of older patients were dry mouth, somnolence, postural hypotension, insomnia, weight gain, and dizziness. Insomnia was reported by >10% of patients receiving placebo.

In a case series of 11 elderly patients with mania receiving asenapine, Baruch et al³⁶ reported a 63% remission rate. One patient discontinued the study because of a new rash, 1 discontinued after developing peripheral edema, and 3 patients reported mild sedation.

Beyer et al³⁷ reported on a post hoc analysis of 94 older adults (mean age, 57.1 years; range, 50.1 to 74.8 years) with acute bipolar mania receiving olanzapine (n = 47), divalproex (n = 31), or placebo (n = 16) in a pooled olanzapine clinical trials database. Patients receiving olanzapine or divalproex had improvement in mania; those receiving placebo did not improve. Safety findings were comparable with reports in younger patients with mania.

Other clinical data. Adverse effects found in mixed-age samples using secondary analyses of clinical trials need to be interpreted with caution because these types of studies usually exclude individuals with significant medical comorbidity. Medical burden, cognitive impairment, or concomitant medications generally necessitate slower drug titration and lower total daily dosing. For example, a secondary analysis of the U.S. National Institute of Health-funded Systematic Treatment Enhancement Program for Bipolar Disorder study, which had broader inclusion criteria than most clinical trials, reported that, although recovery rates in older adults with bipolar disorder were fairly good (78.5%), lower doses of risperidone were used in older vs younger patients.³⁸

Clinical considerations

Interpretation of the relative risks of antipsychotics in older people must be tempered by the caveat that there is limited high-quality data (Table 4). Antipsychotics are the first-line therapy for older patients with schizophrenia, although their use is supported by a small number of prospective RCTs. SGAs are preferred because of their lower propensity to cause EPS and other motor adverse effects. Older persons with schizophrenia have an EPS threshold lower than younger patients and determining the lowest effective dosage may minimize EPS and cognitive adverse effects. As individuals with long-standing schizophrenia get older, their antipsychotic dosages may need to be reduced, and clinicians need to monitor for adverse effects that are more common among older people, such as tardive dyskinesia and metabolic abnormalities. In healthy, “younger” geriatric patients, monitoring for adverse effects may be similar to monitoring of younger patients. Patients who are older or frail may need more frequent assessment.

Like older adults with schizophrenia, geriatric patients with bipolar disorder have reduced drug tolerability and experience more adverse effects than younger patients. There are no prospective controlled studies that evaluated using antipsychotics in older patients with bipolar disorder. In older bipolar patients, the most problematic adverse effects of antipsychotics are akathisia, parkinsonism, other EPS, sedation and dizziness (which may increase fall risk), and GI discomfort. A key tolerability and safety consideration when treating older adults with bipolar disorder is the role of antipsychotics in relation to the use of lithium and mood stabilizers. Some studies have suggested that lithium has neuroprotective effects when used long-term; however, at least 1 report suggested that long-term antipsychotic treatment may be associated with neurodegeneration.³⁹

The literature does not provide strong evidence on the many clinical variations that we see in routine practice settings, such as combinations of drug treatments or drugs prescribed to patients with specific comorbid conditions. There is a need for large cohort studies that monitor treatment course, medical comorbidity, and prognosis. Additionally, well-designed clinical trials such as the DART-AD, which investigated longer-term trajectories of people with dementia taking antipsychotics, should serve as a model for the type of research that is needed to better understand outcome variability among older people with chronic psychotic or bipolar disorders.⁴⁰