Late-onset schizophrenia

Schizophrenia is a heterogeneous clinical syndrome characterized by psychosis that typically begins in late adolescence or early adulthood. The DSM-5 criteria also require functional impairment, but the ICD-11 guidelines do not (DSM-5; ICD-11) (Harrison et al., Reference Harrison, Weber, Jakob and Chute2021). The lifetime prevalence of schizophrenia in the general population is 1% and approximately 20% of older patients with schizophrenia have onset of illness after age 40 (Stafford et al., Reference Stafford, Howard and Kirkbride2018). Late-onset schizophrenia (LOS) with onset between 40 and 60 years of age differs from early-onset schizophrenia (EOS) in several ways. LOS is generally associated with lower severity of positive symptoms (i.e., delusions, hallucinations, disorganized speech and behavior) and lower antipsychotic dose requirement. In 2000, the International Late-Onset Schizophrenia Group proposed the terms LOS for cases with onset between 40 and 60 years of age and very-late-onset schizophrenia-like psychosis (VLOSLP) for those presenting with the first episode of psychosis after age 60 (Howard et al., Reference Howard, Rabins, Seeman and Jeste2000). People who develop LOS typically have better premorbid functioning and a better prognosis than those with EOS. In LOS, persecutory delusions, delusions of reference, and third person and running-commentary auditory hallucinations are more common while disorganized thoughts are less frequent. In a sample of 854 older patients with schizophrenia, including 110 with LOS, patients with LOS had less severe psychotic symptoms, better functioning, and were on lower antipsychotic doses than patients with EOS (Vahia et al., Reference Vahia2010). Women comprise a majority of patients with LOS and VLOSLP with the relative preponderance of women increasing with age (Reynolds et al., Reference Reynolds, Jeste, Sachdev and Blazer2022; Stafford et al., Reference Stafford, Howard and Kirkbride2018). The higher prevalence in women led to the estrogen hypothesis of schizophrenia – i.e., later onset of symptoms in postmenopausal women results from a loss of the previous protection conferred by estrogenic modulation of several neurotransmitters but attempts to identify specific polymorphisms of the estrogen receptors related to LOS have been inconclusive and estrogen therapies have not shown consistent results (Gonzalez-Rodriguez and Seeman, Reference González-Rodríguez and Seeman2019). VLOSLP has some overlapping features with neurodegenerative disorders such as AD and PD with psychosis (Nilsson et al., Reference Nilsson, Sørensen and Enggard2018).

Older patients with schizophrenia, both EOS and LOS, show considerable variability in their outcomes over the course of illness. Some patients experience worsening of psychosis, many have a stable course, and a minority demonstrate progressive improvement in symptoms with partial or full remission (Cohen and Reinhardt, Reference Cohen and Reinhardt2020; Jeste et al., Reference Jeste, Wolkowitz and Palmer2011). A meta-analysis found that LOS patients demonstrated better cognitive performance than typical EOS patients in the domains of memory, executive function, and processing speed, but had more impairment in attention, verbal fluency and visuospatial construction, consistent with specific deficits related to the disorder rather than deficits solely due to aging-related decline (Rajji et al., Reference Rajji, Ismail and Mulsant2009). LOS is not a prodrome of AD because progressive cognitive worsening with an AD cognitive profile is uncommon (Rajji et al., Reference Rajji, Ismail and Mulsant2009). Functional impairments can be severe and disabling even in LOS. Patients with schizophrenia have an average 20% decrease in lifespan compared to the general population; accelerated aging, poor general health and healthcare, and other factors have been postulated to underlie this decrease (Nguyen et al., Reference Nguyen, Eyler and Jeste2018). There has been limited investigation of the genetics of LOS. The genetic variant in the dopamine D2 receptor (DRD2), rs2734839, is reported to be significantly associated with schizophrenia generally, but more with LOS (Voisey et al., Reference Voisey, Swagell, Hughes, Lawford, Young and Morris2012).

Schizophrenia and dementia

Studies of the links between primary psychosis and dementia have focused on schizophrenia, which increases the risk of all-cause dementia (Ribe et al., 2015; Stroup et al., Reference Stroup2021). In a national study from Denmark, schizophrenia increased the risk for all-cause dementia by 2.13-fold (Ribe et al., 2015). A recent study that used data from the US Medicare program suggested an even higher risk (Stroup et al., Reference Stroup2021). At 80 years, about 70% of schizophrenia individuals had a dementia diagnosis compared to 11% in the group without severe mental illness with diagnoses that included AD, senile dementia, and vascular dementia. Increased risk may have been due to some combination of early cognitive impairment, low cognitive reserve, metabolic syndrome, cardiovascular or cerebrovascular disease, neurodegenerative neuropathologic processes, and perhaps long-term use of antipsychotics (Jonas et al., Reference Jonas, Abi-Dargham and Kotov2021; Stroup et al., Reference Stroup2021). However, in these studies, early-onset patients were included and comprised the vast majority of patients with schizophrenia. In another study, postmortem examination of patients with LOS did not show histopathological changes consistent with AD as amyloid plaque burden was minimal (Casanova et al., Reference Casanova, Stevens, Brown, Royston and Bruton2002). Current evidence suggests that LOS is not associated consistently with increased rates of dementia, although dementia is common in VLOSP (Howard et al., Reference Howard, Rabins, Seeman and Jeste2000).

Brain imaging in LOS

For the initial clinical presentation of late-onset psychosis, structural brain imaging with CT or MRI is important to rule out CNS causes such as tumor and stroke. Compared to EOS, LOS has been associated with lateral ventricular enlargement, larger thalamic volumes, decreased white matter integrity, and reduced cerebral blood flow in both postcentral gyri compared to EOS (Chen et al., Reference Chen2013; Tonkonogy and Geller, Reference Tonkonogy and Geller1999; Wake et al., Reference Wake2016).

Principles of antipsychotic prescribing in old age

Pharmacokinetic and pharmacodynamic changes that occur with aging lead to an increased sensitivity to antipsychotics in older individuals. Specifically, decreases in total body water and muscle mass combined with increase in the proportion of adipose tissue result in an increased volume of distribution and slower elimination of antipsychotic medications, while decreased hepatic protein synthesis results in greater amounts of “free” circulating drug (Uchida et al., Reference Uchida, Mamo, Mulsant, Pollock and Kapur2009). Also, increased permeability of the blood–brain barrier with aging can lead to higher brain concentrations of antipsychotic medications. Therefore, older adults are at a greater risk for antipsychotic-induced side effects such as extrapyramidal symptoms (EPS) and tardive dyskinesia (Uchida et al., Reference Uchida, Mamo, Mulsant, Pollock and Kapur2009). There is also an increased risk of falls, autonomic effects including anticholinergic symptoms of dry mouth and constipation, and sedation.

Randomized, double-blind, placebo-controlled clinical trials of antipsychotics in LOS are sparse. In VLOSLP, amisulpride showed moderate efficacy in a clinical trial (Howard et al., Reference Howard2018). In LOS and VLOSLP, the choice of antipsychotic medication is usually determined by the risks of specific side effects, e.g., neurological versus metabolic side effects. Second generation or “atypical” antipsychotics (e.g., risperidone, olanzapine, quetiapine, ziprasidone, aripiprazole, paliperidone, iloperidone, asenapine, lurasidone) are generally associated with a lower risk for parkinsonism and tardive dyskinesia than first generation or “typical” antipsychotics (e.g., chlorpromazine, haloperidol, fluphenazine). Several atypical antipsychotics have elevated risk of metabolic side effects. A study using equipoise-stratified randomization compared four commonly prescribed atypical antipsychotics (aripiprazole, olanzapine, quetiapine, and risperidone) in 332 outpatients over 40 years old with psychotic symptoms related to different psychiatric diagnoses, especially schizophrenia and dementia, over 2 years of treatment (Jin et al., Reference Jin2013). Concerning findings included a high 1-year cumulative incidence of metabolic syndrome (36%), high rates of both serious and nonserious adverse events (50.8%), and no further symptom improvement in these patients, many of whom were already receiving long-term treatment. Over half of the study participants discontinued their assigned medication within 6 months, most often due to side effects (51.6%) or lack of efficacy (26%), suggesting that atypical antipsychotic medications may be helpful short-term but have limited efficacy and an increased risk of adverse effects over longer treatment periods in middle-aged and older adults with schizophrenia. Higher doses of commonly used antipsychotics (risperidone, olanzapine, and haloperidol) have been associated with higher risk of death than lower doses (Gerhard et al., Reference Gerhard2014). In a study of 29 older adults with schizophrenia (early and late onset) whose olanzapine or risperidone dose was reduced by up to 40%, there was no increase in relapse (Graff-Guerrero et al., Reference Graff-Guerrero2015). Clinically, using the lowest effective dose of antipsychotic medications is important to minimize adverse effects.

Nonpharmacological interventions

Cognitive behavioral social skills training, a 36-session weekly group therapy program combining cognitive behavioral therapy, social skills training, and problem-solving training resulted in improved functioning in middle-aged and older patients with schizophrenia or schizoaffective disorder compared with a supportive therapy control (Granholm et al., Reference Granholm, Holden, Link, McQuaid and Jeste2013). In another study of cognitive behavioral social skills training, older patients with schizophrenia who had the greatest executive dysfunction showed the most improvement in independent living skills compared to treatment-as-usual control patients (Rajji et al., Reference Rajji, Mamo and Holden2022). Functional adaptation and skills training, a 24-week functional skills course, was associated with improvement in functioning and decrease in utilization of emergency medical services in older adults with schizophrenia (Patterson et al., Reference Patterson, Mausbach, McKibbin, Goldman, Bucardo and Jeste2006). Both these interventions, which included middle-aged and older adults with EOS and LOS, are considered evidence-based by the Substance Abuse and Mental Health Services Agency (SAMHSA) in the USA. These approaches need greater implementation in clinical practice. Nonpharmacological interventions have not been studied in delusional disorder and psychotic depression.

Delusional disorder

Besides schizophrenia, DSM-5 specifies several psychotic disorders: delusional disorder, brief psychotic disorder, schizophreniform disorder, schizotypal (personality) disorder, schizoaffective disorder, substance/medication-induced psychotic disorder, and psychotic disorder due to another medical condition. Delusional disorder presents in middle to old age with prominent delusions but without hallucinations. Delusional disorder has an estimated prevalence of 0.03% in old age and a lifetime prevalence of approximately 0.2% (Tampi et al., Reference Tampi, Young, Hoq, Resnick and Tampi2019). Women may be more commonly affected than men (Gonzalez-Rodrigues et al., Reference González-Rodrígues2022). A diagnosis of delusional disorder requires ruling out other causes of delusions, including schizophrenia, delirium, neurocognitive disorders, substance-induced disorders, and mood disorders.

Studies of individual cases and others with small sample sizes suggest an association between incident delusional disorder and structural brain changes with cognitive deficits (Gonzales-Rodrigues et al., Reference González-Rodrígues2022). There has been little systematic research on the genetics of delusional disorder. Apart from the direct consequences of delusions, everyday functioning usually is relatively intact (Munoz-Negro et al., Reference Munoz-Negro2018). In delusional disorder, placebo-controlled medication trials have not been conducted. In clinical series, response rates to antipsychotics range from 20 to 39%, and full remission is rarely achieved (Nagendra and Snowdon, Reference Nagendra and Snowdon2020).

Psychotic depression

The estimated lifetime prevalence for psychotic depression is 0.35%, and like delusional disorder, the age of onset of psychotic depression averages around 50 years with no gender difference (Perälä et al., Reference Perälä2007). Unipolar psychotic depression often has an onset in middle to old age, whereas early-onset psychotic depression is more frequent in bipolar disorder (Nelson and Charney, Reference Nelson and Charney1981). Delusions of guilt, worthlessness, nihilism, and somatic delusions are common in psychotic depression, but the subtlety of symptoms and patients’ guardedness may lead to the diagnosis being missed. Delusions are far more common than hallucinations; hence the term delusional depression has been used synonymously with psychotic depression. Auditory hallucinations that are derogatory or belittling, which are mood-congruent with severe depression, can occur. Psychomotor retardation is common and there is greater severity, persistence, and recurrence of episodes than most other forms of depression. Completed suicide is more common in psychotic than nonpsychotic depression (Gournellis et al., Reference Gournellis2018). Patients with psychotic depression show more cognitive deficits in all domains, except for verbal fluency, compared to patients with nonpsychotic depression (Vermeulen et al., Reference Vermeulen, Lauwers, Van Diermen, Sabbe, van der Mast and Giltay2019).

Psychotic depression may be associated with decreased gray matter volume, particularly in medial prefrontal cortex, but there are no consistent MRI changes associated with treatment response to electroconvulsive therapy (ECT) (Takamiya et al., Reference Takamiya2022). In a study that also included patients with schizophrenia and bipolar disorder, no specific genes were associated with psychotic depression (Luna et al., Reference Luna2022).

Treatment with combined antipsychotic and antidepressant medications is the usual first-line strategy. In a randomized, double-blind, controlled trial (RCT) of 259 patients with psychotic depression, two-thirds of whom were older adults, a combination of olanzapine and sertraline was associated with a 41.9% remission rate compared to 23.9% remission rate on olanzapine plus placebo (Meyers et al., Reference Meyers2010). ECT, however, shows the strongest efficacy in this disorder in all age groups, including old age, and complete remission can be achieved (Devanand and Krueger, Reference Devanand and Krueger1994). Therefore, in patients showing partial or no response to combination pharmacotherapy, ECT is recommended.

Bipolar disorder with psychosis

Bipolar disorder appears to have a trimodal age-of-onset distribution—early-onset with modal age of 17, mid-onset with modal age of 26, and late-onset with modal age of 46 years (Bolton et al., Reference Bolton, Warner, Harriss, Geddes and Saunders2021). Approximately 25% of people with bipolar disorder are older than 60 years (Sajatovic et al., Reference Sajatovic2015). Older adults with bipolar disorder may experience psychotic symptoms in depressive or manic episodes (Bolton et al., Reference Bolton, Warner, Harriss, Geddes and Saunders2021). In older adults, new-onset bipolar disorder is not driven primarily by genetic predisposition, unlike early-onset bipolar disorder (Sajatovic et al., Reference Sajatovic2015). The term “secondary mania” is used to describe mania that is related to medication toxicity, infections, and metabolic changes as well as brain pathology due to brain trauma and neurological illnesses such as stroke (Evans et al., Reference Evans, Byerly and Greer1995). Secondary mania is more common in older than in younger adults. Psychotic symptoms in secondary mania are similar to those in early-onset bipolar disorder. Since lithium may lead to cognitive impairment in older patients with neurodegenerative disease, accurate diagnosis is important.

First-line therapy in older adults is comparable to treatment for younger adults with bipolar disorder, with the caveat that doses need to be lower because of increased age-related risk for neurological and metabolic side effects of lithium, anticonvulsants, and antipsychotics. Compared to younger adults, older adults require a lower dose of lithium to achieve a therapeutic plasma level, because of age-related decline in renal function and excretion of lithium (Malhi et al., Reference Malhi, Gessler and Outhred2017; Dols and Beekman, Reference Dols and Beekman2018). A double-blind RCT of lithium and divalproex in older adults with mania, with and without psychosis, found no significant differences in efficacy between the two treatments (Young et al., Reference Young2017). Clinically, lithium or anticonvulsants like valproate are effective treatments, particularly as maintenance treatments, with acute manic episodes often requiring the addition of antipsychotics. Most studies of bipolar disorder have been conducted in young to middle-aged adults. More research on the prevalence, etiology, treatment, and clinical course of late-onset bipolar disorder, with and without psychosis, is needed (Sajatovic et al., Reference Sajatovic2015).

Isolated hallucinations in older adults

Isolated auditory or visual hallucinations occur in 2–3% of older adults (Badcock et al., Reference Badcock, Dehon and Larøi2017). A range of predisposing factors include decline in sensory or cognitive functioning, poor sleep, and psychosocial stressors that include social isolation, loneliness, and bereavement. There is no established treatment for idiopathic isolated hallucinations, and antipsychotics often are ineffective (Badcock et al., Reference Badcock, Dehon and Larøi2017). In many cases, it is unclear if the hallucination is a result of abnormal functioning of the peripheral sensory pathways for hearing and vision that cannot be detected by testing, or due to abnormal brain physiology. In the presence of Parkinsonian features, however, the diagnosis of LBD leading to dementia needs to be considered. Few genetic, brain imaging, and treatment studies of this uncommon disorder have been conducted (Badcock et al., Reference Badcock, Dehon and Larøi2017).

Secondary psychoses

Secondary psychoses involve a specific etiology for psychosis that may or may not be reversible (Keshavan and Kaneko, Reference Keshavan and Kaneko2013). Traumatic brain injury, autoimmune disorders, stroke, CNS malignancies including temporal lobe tumors, and CNS infections can be associated with psychotic features. Other causes of secondary psychosis that can occur throughout the life span include seizure disorders, endocrine disorders, metabolic disorders, and drug use. Illicit drug use can lead to specific psychotic symptoms based on the individual drug or combination of drugs, but these occur primarily in younger or middle-aged adults. Secondary psychoses are often characterized by visual hallucinations, disorientation/confusion, and a medical illness with onset of psychosis shortly after a new medical diagnosis or medication. Treatment involves addressing the cause of the secondary psychosis, which in some cases may be reversible. The remainder of this review focuses on common causes of secondary psychosis.

Drug-induced toxicity and psychosis

Illicit drugs are primarily associated with psychosis in adolescents and young adults but occur occasionally in older adults who use them. Stimulants like amphetamines and cocaine, and psychotomimetic drugs like phencyclidine and ketamine, can induce psychosis (Keshavan and Kaneko, Reference Keshavan and Kaneko2013). Withdrawal from alcohol and benzodiazepines can be associated with psychosis. Lysergic acid diethylamide and 3,4-methylenediozxy-N-methylamphetamine can lead to hallucinations during acute intoxication but these symptoms do not persist in the abstinent state. Opioids typically do not induce psychosis.

Prescription drugs can cause psychosis in older adults. Corticosteroids can induce depression, mania and psychosis, and medications with anticholinergic properties can be associated with psychosis as an acute delirium or a low-grade chronic delirium. Medications with anticholinergic properties include tricyclic antidepressants and older antiparkinsonian agents. There are case reports and clinical series of treatment with anticonvulsants, antiemetics, histamine antagonists, antimalarials, and antibiotics being associated with psychosis (Keshavan and Kaneko, Reference Keshavan and Kaneko2013). The management of drug-induced psychosis consists of identifying and then tapering or discontinuing the drug that likely induced the psychosis.

Delirium

In older adults, delirium and neurodegenerative disorders are common causes of secondary psychosis (Table 1). The DSM-5 criteria for delirium emphasize impaired attention and awareness, abrupt onset with fluctuating course, and at least one type of cognitive impairment (e.g., memory, language, orientation). The DSM-5 diagnostic criteria for delirium include a criterion for perceptual disturbance, which subsumes hallucinations.

Table 1. Characteristic features of common late-onset psychotic disorders

EOS: early onset schizophrenia. CBT: cognitive behavioral therapy. RCT: randomized clinical trial. ECT: electroconvulsive therapy. REM: rapid eye movement. PET: positron emission tomography. EPS: extrapyramidal signs.

There are multiple risk factors for delirium including age, alcohol and drug abuse, pre-existing cognitive compromises, medication toxicity, and surgery (Rengel et al., Reference Rengel2021)). Delirium is associated with need for greater care, longer hospital stays, increased hospitalization costs, and increase in mortality (Inouye et al., Reference Inouye, Westendorp and Saczynski2014). In intensive care units, rates of delirium have been reduced to about 16–33% from a high incidence of 80% in the early 2000s (Hayhurst et al., Reference Hayhurst, Pandharipande and Hughes2016). We showed in a meta-analysis of 24 studies that delirium was associated with long-term cognitive decline with a Hedges G medium effect size of 0.45 (Goldberg et al., Reference Goldberg2020). The pathophysiology of delirium is not established but it may involve some combination of neuroinflammation, abnormalities in multiple neurotransmitters, and physiological effects of surgery/general anesthesia.

In an early study of 227 consecutively hospitalized patients diagnosed with delirium using DSM-IV criteria, the prevalence of psychotic symptoms was 42.7%, with 27% of patients having visual hallucinations, 12% having auditory hallucinations, 3% having tactile hallucinations, and 26% having delusions (Webster and Holroyd, Reference Webster and Holroyd2000). Visual hallucinations, but not delusions or auditory hallucinations, were associated with more medical diagnoses and multiple etiologies. In another study using DSM-IV criteria and a delirium rating scale, psychotic symptoms were observed in 49% of the sample of over 200 patients (Paik et al., Reference Paik, Ahn, Min, Park and Kim2018). Psychosis was associated with a significant increase in in-hospital mortality, the hyperactive delirium subtype, and unsurprisingly, antipsychotic use. A Cochrane review of nine studies found that antipsychotics did not reduce delirium severity compared to nonantipsychotic drugs or placebo, and antipsychotics also did not change the mortality rate (Wu et al., Reference Wu2019).

In summary, delirium is common in hospitalized older patients and the presence of psychosis is associated with poor prognosis. Besides reversing the cause of delirium when feasible, commonly used medication strategies that include antipsychotics are not supported by a robust evidence base. Individualized clinician decision-making, along with environmental strategies, is the norm.

Major neurocognitive disorders

In the USA, AD comprises more than half of all cases of dementia, vascular dementia and Lewy body dementia (LBD) each contribute to 10–25% of cases of dementia, and frontotemporal dementia (FTD) comprises 2–10% of cases of dementia (Devanand et al., Reference Devanand, Lee, Huey and Goldberg2022; Naasan et al., Reference Naasan2021). Neuropathological studies indicate that overlap is common with more than one type of dementia identified in individual brains at autopsy (Devanand et al., Reference Devanand, Lee, Huey and Goldberg2022).

The diagnostic criteria for psychosis in AD and related dementias require the presence of delusions or hallucinations, absence of delirium or a prior functional psychosis, and onset of psychosis after the clinical onset of dementia (Fischer et al., Reference Fischer2020; Jeste and Finkel, Reference Jeste and Finkel2000). Psychosis in dementia heralds a poor prognosis with greater cognitive decline and increased risk of institutionalization and death (Scarmeas et al., Reference Scarmeas2005). Psychotic symptoms are often accompanied by agitation and aggression with markedly increased burden on caregivers and the healthcare system.

The prevalence of psychosis, defined by the presence of delusions or hallucinations, differs by clinically diagnosed subtype of dementia: 10–15% in behavioral variant of FTD (bvFTD) with a similar prevalence in vascular dementia, 30–50% in AD, and 30–70% in dementia with Lewy bodies (Tampi and Jeste, Reference Tampi and Jeste2022). These estimates reflect prevalence over the entire disease course; prevalence for a specific timepoint or disease stage is lower because psychosis typically fluctuates over time in major neurocognitive disorders.

Alzheimer’s disease (AD)

Psychosis occurs in 30–50% of patients with AD during their disease course. Delusions occur in approximately 15–30% of patients with mild to moderate AD without any increase in prevalence with disease severity (Devanand et al., Reference Devanand1997; Lyketsos et al., Reference Lyketsos, Lopez, Jones, Fitzpatrick, Breitner and DeKosky2002). Delusions of theft are common and typically directed at the spouse or other caregivers; delusions of persecution, infidelity, and abandonment also occur. Phantom boarder syndrome, which is the belief that someone else is staying in the home, is less common but can be disruptive to the patient and the caregiver. Misidentification due to cognitive worsening can manifest as a delusion, including the belief that one’s house is not one’s home or that a family member is an imposter (Capgras syndrome). Hallucinations occur in 5–15% of patients in the mild to moderate stage, may not be well-defined, and manifest in any sensory modality with visual and auditory hallucinations being the most common. Agitation is commonly comorbid with psychosis.

Neurodegeneration in most types of dementia, particularly AD and LBD, is associated with degeneration of the major neurotransmitter pathways, including dopaminergic pathways. There may be a therapeutic window of D2/D3 receptor occupancy to treat psychosis in AD (Reeves et al., Reference Reeves2017). Serotonin (5-HT) is reduced in the ventral temporal cortex and prosubiculum in AD with psychosis compared to AD without psychosis, which may be related to lower cell counts in the dorsal raphe nucleus in AD with psychosis (Creese et al., Reference Creese, Ballard, Aarsland, Londos, Sharp and Jones2014). There may be an altered monoamine-cholinergic balance in both AD and LBD patients with psychosis (Perry et al., Reference Perry1990). Increased neocortical neurofibrillary tangle density has been reported in AD with psychosis (Farber et al., Reference Farber2000). In AD psychosis, PET studies have been limited while MRI studies show associations of delusions with increased frontotemporal, hippocampal, and parahippocampal atrophy (Lee et al., Reference Lee2019). The risk for psychosis in AD may be familial with an estimated heritability of 61%, which compares with 50–60% for AD more broadly (Hollingworth et al., Reference Hollingworth2012).

Behavioral interventions that include caregiver education are recommended as first-line treatment based on small to medium effect sizes in single-blind trials of agitation in dementia (Tampi and Jeste, Reference Tampi and Jeste2022). Psychosis, which often accompanies agitation, has not been the primary outcome in these behavioral trials.

Pharmacological treatments currently are not approved for psychosis in patients with dementia in the USA, and only a few countries have approved the use of atypical antipsychotics in limited circumstances in patients with dementia. Nevertheless, many patients are treated off-label with antipsychotics; placebo-controlled RCTs of these medications have shown inconsistent efficacy with small effect sizes. Among the atypical antipsychotics, risperidone in the dose range of 0.5–2 mg daily has the strongest evidence for efficacy with a small to medium effect size (Katz et al., Reference Katz, de Deyn, Mintzer, Greenspan, Zhu and Brodaty2007). Antipsychotic use in patients with dementia is associated with increased risk of mortality with an odds ratio of 1.6 that led the FDA to issue a boxed warning in 2005 (Schneider et al., Reference Schneider, Dagerman and Insel2005). Antipsychotic use in older patients with dementia and psychosis is also associated with other adverse reactions including EPS, autonomic side effects, and cognitive impairment. Both RCTs and electronic health record studies show that the risk of adverse effects and mortality increases markedly for higher compared to lower doses for all commonly used antipsychotics (Gerhard et al., Reference Gerhard2014; Huybrechts et al., Reference Huybrechts2012). In the large NIMH-sponsored Clinical Antipsychotic Trials of Intervention Effectiveness–Alzheimer’s Disease (CATIE-AD) trial, which compared olanzapine, quetiapine, and risperidone to placebo for psychosis, aggression, and agitation in AD, the benefits of antipsychotic treatment on symptoms were offset by adverse effects (Schneider et al., Reference Schneider2006).

The advent of atypical antipsychotics led to a decrease in tardive dyskinesia in older adults, though EPS remain a concern with increased risk of falls. Recently, the FDA in the USA has approved two medications for treatment of tardive dyskinesia – viz., valbenazine and deutetrabenazine (McEvoy, Reference McEvoy2019). Standardized measures to assess behavioral symptoms in neurocognitive disorders, such as the Neuropsychiatric Inventory, can help to monitor the effects of treatment. Antipsychotic discontinuation in patients with dementia in nursing homes, which is required by the Centers for Medicare and Medicaid Services in the USA “unless clinically contraindicated”, is associated with an increased rate of relapse; this risk of relapse needs to be balanced against the increased likelihood of adverse effects with long-term antipsychotic treatment (Devanand et al., Reference Devanand2012). Although very few countries have approved the use of antipsychotics to treat psychosis in patients with dementia, based on the available evidence, antipsychotics may be used in individuals with psychosis associated with dementia when the symptoms are severe or refractory and when nonpharmacological treatments have not produced benefit or are not safe or feasible to employ. Furthermore, antipsychotics should be used at the lowest effective dosage and for the shortest possible time, with close monitoring of risk factors and adverse effects (Tampi and Jeste, Reference Tampi and Jeste2022).

Lewy body dementia (LBD)

LBD symptoms include progressive memory loss, visual hallucinations, parkinsonism, cognitive fluctuations, and rapid eye movement sleep behavior disorder. LBD is associated with psychosis in 30–80% of patients during the disease course. Hallucinations are more common in LBD than AD. Visual hallucinations are one of the criteria for making the diagnosis of LBD but not all patients with LBD manifest this symptom (McKeith et al., Reference Mckeith2017). Hallucinations in LBD range from incompletely formed passage hallucinations (perception in the peripheral visual field that a person or animal is passing by) to presence phenomena (someone is present even though no one is there) (Outeiro et al., Reference Outeiro2019). Pareidolias, which are complex visual illusions involving ambiguous forms that are perceived as meaningful objects, are analogous to visual hallucinations in LBD.

In clinically diagnosed patients with LBD, SPECT, or PET scan shows decreased dopamine transporter as in PD, and CSF biomarkers for AD may be positive in LBD (Outeiro et al., Reference Outeiro2019). In patients with dementia who manifest psychosis during life, AD alone, LBD alone, and comorbid AD with LBD are all commonly identified in autopsied brains (Devanand et al., Reference Devanand, Lee, Huey and Goldberg2022).

Patients with LBD are more sensitive to antipsychotic medications and at high risk for severe adverse reactions including EPS and anticholinergic and hypotensive effects. In small clinical trials, low doses of olanzapine and quetiapine showed limited efficacy in the treatment of psychosis in LBD (Cummings et al., Reference Cummings, Street, Masterman and Clark2002; Culo et al., Reference Culo2010).

Cognition and mechanisms of psychosis in AD and LBD

The cognitive profile of psychosis in AD and LBD may be a more severe variant of a modal neurocognitive profile in which episodic memory disturbances are most prominent, along with slowing in speed of processing and impairments in semantic, naming, and executive abilities, as well as visual processing/visual construction impairments. These impairments reflect the progression of neurodegeneration but are not specific to psychosis.

In a series of 1808 brain autopsies from the National Alzheimer’s Coordinating Center (NACC) database in the USA, we found an association between greater severity of psychotic symptoms and greater evidence of pathology in AD (Devanand et al., Reference Devanand, Lee, Huey and Goldberg2022). Increase in alpha-synuclein pathology, which indicates LBD, from midbrain to limbic and then neocortical areas was strongly associated with the presence and severity of psychosis. Neocortical Lewy bodies are prominent in parietal and occipital cortex but interestingly, not primary visual cortex. There are, however, severe impairments in visual and visual spatial identification in such patients, and these deficits may promote visual hallucinations or pareidolias. In PET studies of LBD psychosis, misidentification syndromes were associated with insula, prefrontal, and hippocampal hypoperfusion, perhaps consistent with a loss of interoceptive or mnemonic confirmation of identity, along with a failure in monitoring reality. Visual hallucinations in LBD correlate with lower glucose metabolism and weaker metabolic connectivity in the parietal-occipital cortex, but stronger connectivity in the insula and prefrontal cortex (Ffytche et al., Reference Ffytche2017). In AD, both the extra-striatal dorsal and ventral visual streams for visual processing are heavily compromised and could result in bottom-up-driven visual hallucinations. In particular, the fusiform face area and parahippocampal ‘place’ area appear to be involved. Visual hallucinations are more frequent than auditory hallucinations in AD and LBD in contrast to predominantly auditory hallucinations in schizophrenia, suggesting different underlying mechanisms.

Psychosis in FTD

There are several subtypes of FTD, most of which have specific underlying genetic abnormalities. In patients with FTD, the neuropathological diagnosis is termed frontotemporal lobar degeneration (FTLD). Description of the genetics of the large number of genetic subtypes in FTD is beyond the scope of this review. Delusions and hallucinations are less common in FTD and vascular dementia compared to AD and DLB. In an autopsy study, hallucinations were uncommon in patients with FTD but delusions, including paranoid and somatic delusions, occurred in up to one-third of patients with the TDP-43 subtype but was rare in the tau-subtype of FTD (Naasan et al., Reference Naasan2021). In the NACC consortium postmortem series of 1808 brains with dementia diagnoses, psychotic symptoms were present in 22% of FTLD cases with a higher prevalence in AD and LBD (Devanand et al., Reference Devanand, Lee, Huey and Goldberg2022). There have been no clinical trials to treat psychosis in FTD and antipsychotics are rarely used clinically in this condition.

Vascular dementia

Vascular dementia may be a consequence of a large stroke, multiple lacunes and infarcts, or extensive microvascular pathology. While psychosis has been described in vascular dementia, it is less common than in AD or LBD and psychosis in these patients may be related to comorbid AD or LBD. Patients with vascular dementia and mixed dementia (vascular dementia with AD) have been included in antipsychotic trials that focused on patients with AD. These trials showed no differences in efficacy between AD and mixed dementia (Mühlbauer et al., Reference Mühlbauer, Möhler, Dichter, Zuidema, Köpke and Luijendijk2021).

PD psychosis

PD is a disease of the basal ganglia resulting in abnormalities in motor function, and psychosis can develop during the illness. During the course of illness, psychotic symptoms occur in 60% of patients with PD, often because of treatment with excessive doses of dopaminergic medications, including the widely used L-DOPA Carbidopa combination that targets the basal ganglia dopamine deficit (Ffytche et al., Reference Ffytche2017). Lowering the doses of these drugs to reduce psychotic symptoms usually worsens rigidity and tremor, leading to difficulty in balancing efficacy against side effects of dopaminergic anti-Parkinsonian drugs. Hallucinations are prominent in PD psychosis. Some drug-naïve patients, however, may experience “minor” hallucinations and correlations between the dose of the dopamine agonist and hallucinations are not strong (Lenka et al., Reference Lenka, Pagonabarraga, Pal, Bejr-Kasem and Kulisvesky2019). Hallucinations have been associated with rapid eye movement (REM) sleep behavior disorder in PD, indicating possible LBD in the neocortex. Quetiapine in low doses is used clinically to treat psychosis in PD but controlled evidence of its efficacy is lacking. Pimavanserin has been approved for the treatment of psychosis in PD in the USA.

Table 1 describes the key commonalties and distinctions among late-onset psychotic disorders in older adults. In the table, we have included common disorders reviewed in this paper that need consideration in the differential diagnosis of new-onset psychosis in older adults.