The off-label use of conventional and atypical antipsychotics to treat the behavioral and psychological symptoms in dementia remains common despite their limited efficacy. Despite their perceived better safety, some atypical antipsychotics may still possess anticholinergic and hypotensive effects, as well as extrapyramidal effects including tardive dyskinesias, although at a lower rate than conventional antipsychotics.

In April 2005, the US Food and Drug Administration (FDA) issued a public health advisory that the use of atypical antipsychotics such as risperidone, olanzapine, quetiapine, and aripiprazole to treat older adults with dementia was associated with an increased risk for death compared with placebo. In a pooled analysis of trials involving atypical antipsychotics, the mortality rate was 60–70 % higher than with placebo in 15 of the 17 trials. Evidence providing support for these warnings has also raised further safety concerns about conventional antipsychotics (Scneider et al. 2005; Wang et al. 2005; Kryzhanovskaya et al. 2006). In June 2008, the FDA stated that the conventional antipsychotics share a similar or even higher risk of increased mortality with the atypical antipsychotics. The FDA concluded that antipsychotics should not be used for the treatment of dementia-related psychosis. Despite these warnings, a recent review suggested that antipsychotics should be “used with caution only when non-pharmacologic approaches have failed to adequately control behavioral and psychological symptoms in dementia” (Trifirò et al. 2009).

Antipsychotics appear to confer a greater mortality risk in older adults than any other psychotropic medication. A retrospective case-control study of 90,786 dementia patients within the Veterans Health Administration from 1998 to 2009 found that newly prescribed antipsychotic users had an increased mortality risk when compared to both matched nonusers and antidepressant users (Maust et al. 2015). This risk ranged from 2.0 % for those receiving quetiapine to 3.8 % for those receiving haloperidol when comparing antipsychotic users to nonusers. When comparing antipsychotic users to antidepressant users, the mortality risk ranged from 3.2 % for those receiving quetiapine to 12.3 % for those receiving haloperidol. Additionally, a dose-response increase in mortality risk was observed for atypical antipsychotics (olanzapine, quetiapine, and risperidone).

The potential causes of death associated with antipsychotic use merit consideration of the potential for drug-drug and drug-disease interactions. Several plausible mechanisms can be proposed, including cerebrovascular events, arrhythmias and sudden cardiac death, venous thromboembolism and pulmonary embolism, and aspiration pneumonia. Metabolic effects of antipsychotics may be a longer-term safety concern. Antipsychotics may contribute to events that are not initially recognized as the first step in a sequence that promotes premature death, such as falls leading to hip fractures. Apart from mortality, other serious safety issues have been raised regarding the use of antipsychotics including cerebrovascular events, cardiac events, peripheral vascular disease, metabolic disease, infections, and falls/fractures. Hip fracture, stroke, myocardial infarction, and ventricular arrhythmias partially explain the mortality difference between first- and second-generation antipsychotics (Jackson et al. 2014).

A pooled analysis documented a threefold increased risk of stroke and transient ischemic attacks for risperidone and olanzapine compared with placebo (De Deyn et al. 2005). Following the warning on stroke and antipsychotic use, observational studies compared the risk of stroke between atypical and conventional antipsychotics, most in the general elderly population and in patients with dementia. The studies suggest no increased risk of stroke with atypical compared to conventional antipsychotics but could not explore the dose and duration effect and risks of individual compounds or rule out confounding by indication because of the strong interrelationship between ischemia, dementia, and subsequent strokes. Stroke may be related to the first weeks of treatment (Sacchetti et al. 2010).

With respect to cardiac events, since the 1960s, sudden cardiac death has been reported with conventional antipsychotic use (Straus et al. 2004) because of their ability to prolong the QTc interval which may result in torsade de pointes and other ventricular arrhythmias (Reilly et al. 2000). Observational studies have confirmed the signals from spontaneous reports, and suggested conventional antipsychotics are associated with an increased risk of sudden cardiac death. A recent systematic review of the association between antipsychotics and myocardial infarction was inconclusive owing to the heterogeneity of the studies, but the largest study revealed no association (Brauer et al. 2011).

A relationship between antipsychotics and venous thromboembolism (VTE) was first suggested five decades ago. Reviews of the available data for aripiprazole, clozapine, and olanzapine have led to warnings about VTE being added to their Summaries of Product Characteristics (MPA Report 2009). The studies on VTE and antipsychotics mostly focus on schizophrenia or young patients and have methodological issues such as small sample sizes and inadequate control of confounding. Inconsistencies in findings have been noted, but an increased risk of VTE is likely (Zhang et al. 2011). Little data are available on the peripheral vascular effects of antipsychotics in dementia, which is highly relevant given the large potential for interacting comedication on serotonin receptors and platelet function.

In patients with either schizophrenia or bipolar disorder, atypical antipsychotics such as olanzapine, quetiapine, and clozapine are associated with metabolic abnormalities including weight gain, lipid disturbances, and altered glucose homeostasis, known risk factors for cardiovascular events (Newcomer 2005). Whether older adults with behavioral and psychological symptoms in dementia receiving antipsychotics develop similar disturbances is controversial as food intake is reduced in these patients. Only a few small studies have been published so far on this association. A study of 36 residents showed treatment with low-dose atypical antipsychotics did not lead to weight gain or increased risk of developing type 2 diabetes or lipid metabolism abnormalities (Rondanelli et al. 2006). In contrast, CATIE-AD reported weight gain but no effect on glucose, total cholesterol, or triglyceride levels during the use of olanzapine, quetiapine, and risperidone, and the risk increased over time. Post-hoc analyses of other studies with olanzapine and risperidone were consistent with the CATIE-AD results. A recently published Canadian study found that among older patients with diabetes, the initiation of treatment with antipsychotic drugs was associated with an increased risk of hospitalization for hyperglycemia (Lipscombe et al. 2009). The risk was high during the initial course of treatment and was increased with the use of all antipsychotic agents. Among nursing home residents with dementia, conventional antipsychotics particularly short-term therapy, but not atypical antipsychotic use, increased risk of diabetes onset (Jalbert et al. 2011).

Infections, primarily pneumonia, have been listed as one of the most prevalent causes of death among demented older adults using antipsychotics both in clinical trials and observational studies. Although one study reported a threefold increased risk with atypical antipsychotics and a 1.6-fold increase with conventional antipsychotics compared with nonuse (Knol et al. 2008), others found a slightly higher rate of fatal pneumonia during conventional antipsychotic use relative to atypical antipsychotic use. The overall risk of antipsychotic use was not increased compared to nonuse in a cohort of elderly persons (Setoguchi et al. 2008). Trifirò showed that the use of either atypical or typical antipsychotics in older patients is associated in a dose-dependent fashion with the development of community-acquired pneumonia (Trifirò et al. 2010). More work is needed to understand this effect (Trifirò 2011).

Information regarding adverse effects of antipsychotic treatment in older adults with schizophrenia is scant. While older adults with schizophrenia are thought to have a greater sensitivity to treatment-related adverse effects (Masand 2000), the overall incidence of adverse events was low (Lasser et al. 2004).

In general, antidepressants are equally effective in the treatment of depression (Mottram et al. 2006). The major categories of antidepressants and individual drugs differ in their side effect and drug interaction profiles. Although for younger patients evidence exists that suggests the use of antidepressants may increase the risk of suicidality and suicidal ideation, among older adults with depression, the risk is reduced. Selective serotonin reuptake inhibitors (SSRIs) are the most commonly prescribed antidepressant class including in frail older adults. Their widespread use has been based primarily on their lack of traditional tricyclic antidepressant (TCA) side effects, although SSRIs commonly produce adverse gastrointestinal and sexual symptoms and less frequently central nervous system effects including insomnia, anxiety, and tremors (Grimsley and Jann 1992).

Both tertiary and secondary amine TCAs have been used in older adults for many years. TCA side effects commonly include dry mouth, blurred vision, urinary retention, and constipation due to their anticholinergic and antihistaminic activity. Amoxapine has been associated with a higher risk of extrapyramidal side effects including akathisia and tardive dyskinesia due to its metabolism to loxapine, a neuroleptic, and maprotiline with seizures (Rosenstein 1993). Frail older adults remain at the greatest risk of anticholinergic effects, cardiovascular effects, and effects on appetite. The anticholinergic effects of older TCAs have included the loss of accommodation with blurring of vision, dry mouth, constipation, urinary retention, tachycardia, confusion, and delirium (Cole and Bodkin 1990). These effects range from what might be considered minor in nature (i.e., blurring of vision and dry mouth), to moderate (i.e., urinary retention), to potentially very serious reactions (i.e., delirium). Anticholinergic side effects minor in nature might potentially have important effects on the older adult’s quality of life. As an example, while tolerance to some anticholinergic effects is known to occur, the loss of accommodation generally does not improve over time and would limit the individual’s ability to perform the simple pleasurable act of reading a book. The presence of concomitant diseases such as prostatic hypertrophy or diabetes mellitus, for example, in older adults may increase the risk that clinically important urinary retention will occur resulting in detrimental effects on the individual’s continence and overall functional status. Most importantly, confusion and delirium may develop from the use of strongly anticholinergic drugs such as amitriptyline in individuals with preexisting cognitive impairment. Potentially, the resulting problems with memory, concentration, and behavioral disturbances may be overlooked as part of the patient’s primary illness or inappropriately treated with a neuroleptic such as haloperidol or with physical restraints. Trazodone has sedating properties (Nierenberg 1994).

Cardiovascular side effects of TCAs must also be considered. Orthostatic hypotension and cardiac conduction defects are the most common TCA cardiovascular side effects. Orthostatic hypotension is due to the blockade of alpha-1 adrenergic receptors. When orthostatic hypotension is symptomatic and results in syncope, older adults are at increased risk of falls and fractures. Although symptomatic orthostatic hypotension can be minimized by the use of small doses of secondary amines and adequate ambulation and hydration, the risk of this side effect remains. TCAs may induce heart block in individuals with preexisting conduction delays.

TCAs may increase appetite and weight, potentially due to their effects on histaminergic systems. Although viewed as a negative outcome in younger populations, weight gain may have beneficial effects in older adults if their nutritional status is poor or marginal due to decreased appetite.

SSRIs other than paroxetine generally lack the anticholinergic properties associated with TCAs. With respect to effects on appetite and weight, SSRIs have been associated with either no change or a decrease in weight at least in the short term (Kinney-Parker 1988). The effects of SSRIs on the cardiovascular system are controversial with citalopram associated with QT prolongation. In one clinical trial for treating agitation in patients with probable Alzheimer’s disease, citalopram titrated from 10 to 30 mg daily was associated with significant improvements in agitation but also with an increase in QTc interval when compared to placebo (Porsteinsson et al. 2014). Worsened cognition was also seen in this trial. Current recommendations are to limit the dosage of citalopram to no more than 20 mg daily in people over 60 years of age. In older adults, multiple risk factors for torsade de pointes are frequently present including hypokalemia, hypomagnesemia, and bradyarrhythmias. SSRIs have been associated with an increased risk of bleeding episodes especially among older adults. Concomitant use of antiplatelet or anticoagulant drugs for atrial fibrillation or myocardial infarction in older adults likely increases the underlying risk (Jiang et al. 2015). Although hyponatremia has been reported to occur in an estimated 10 % of older adults treated with antidepressants (Mannesse 2013), the risk may be greater than SSRIs even considering their widespread usage. In addition, serotonin syndrome presenting as restlessness, anxiety, agitation, and confusion in older adults has been associated with the use of SSRIs and SNRIs. This side effect has been reported to occur primarily when the antidepressant was used in combination with other drugs which have an effect on serotonin such as buspirone, tramadol, and dextromethorphan.