Antipsychotics were advanced for treating agitation, hostility, and aggression in elderly patients since their advent in the 1950s. Chlorpromazine was advertised “for prompt control of senile agitation” (
1). The basis for this claim was case series, newspaper articles featuring physicians extolling chlorpromazine’s virtues for patients in nursing homes (
2), and medical pamphlets promoted by the pharmaceutical company Smith, Kline and French Laboratories (
1).
The 1962 Kefauver-Harris amendment to the 1938 Food, Drug, and Cosmetic Act required, for the first time, that pharmaceutical companies prove that their medications were safe and effective for their intended purpose based on “adequate and well-controlled studies,” and obtain approval from the U.S. Food and Drug Administration (FDA) before marketing the drugs. Despite the Kefauver-Harris restrictions, the use of subsequently marketed antipsychotics for elderly patients, most with dementia, caught on, and the drugs were advertised for this nonapproved purpose, especially thioridazine and haloperidol.
In 1987 Carl Salzman reviewed the treatment of agitation in elderly patients for the American College of Neuropsychopharmacology’s textbook and identified 69 studies of antipsychotics used mostly with dementia (
3). About half were controlled studies, usually comparing one antipsychotic with another, and most were 6 weeks or less in duration. Salzman concluded that neuroleptics have consistent, reliable, but modest therapeutic effects for agitation, and no particular neuroleptic was more advantageous than another. Other reviewers at the time similarly found few controlled trials of neuroleptics that showed positive effects, and some showed worsening; there was weak evidence to support their use (
4,
5). A contemporaneous meta-analysis reported only seven randomized placebo-controlled trials of adequate quality, none statistically significant; but when the trials were statistically combined, the findings suggested that 18 out of 100 patients with dementia would benefit from a neuroleptic over and above treatment with placebo (
6). Adverse events and safety were not taken into account by the reviewers.
The Omnibus Reconciliation Act of 1987, implemented in 1990 after nearly three decades of perceived misuse of neuroleptics as “chemical restraints,” provided new regulations for oversight and monitoring of nursing homes, ensuring an appropriate use of neuroleptics and physical restraints, which practically meant reducing the use of antipsychotics (
7).
The second-generation or “atypical” antipsychotics, including risperidone, olanzapine, and quetiapine, were approved by the FDA for schizophrenia and acute mania in the mid-1990s. The “atypical” designation was earned because of less D
2 receptor blocking and more 5-HT
2 subtype receptor antagonism than many older antipsychotics. Less akinesia and parkinsonian signs and symptoms with atypical antipsychotics was perceived as a relative safety advantage over earlier antipsychotics. The newer antipsychotics were widely considered to be potentially safer than the older ones that were under scrutiny in nursing homes. Large randomized placebo-controlled trials for dementia with agitation, aggression, or psychosis were undertaken by pharmaceutical companies hoping for FDA marketing approval (
8). The studies generally showed small statistical effects on clinical scales, with dropouts and adverse events further limiting effectiveness (
8). In a comparative effectiveness trial of three atypical antipsychotics, adverse effects tended to offset any advantages in the efficacy of treating agitation, psychosis, and aggression in patients with Alzheimer’s disease (
9).
By the early 2000s, the FDA began to share their concern about adverse cardiovascular events, stroke, and deaths occurring in people with dementia participating in atypical antipsychotic trials, generally at annual meetings of the National Institute of Mental Health’s New Clinical Drug Evaluation Unit (NCDEU) (
10–
15). Warnings added to prescribing labels about cardiovascular adverse events followed, as well as the boxed warning in 2005 indicating that “elderly patients with dementia-related psychosis treated with antipsychotic drugs are at an increased risk of death” and that antipsychotics “are not approved for use in patients with dementia-related psychosis” (
16).
In this issue of the
Journal, Gerlach and colleagues, from the University of Michigan and the VA Ann Arbor Healthcare System, tracked prescriptions of psychotropic drugs for veterans with dementia residing in Veterans Health Administration (VHA) nursing homes (
17). During a 10-year period beginning in 2009, the VHA launched the Psychotropic Drug Safety Initiative (PDSI), which aimed, in part, to reduce the proportion of antipsychotics and benzodiazepines prescribed to veterans with dementia. Antipsychotic and benzodiazepine prescribing rates were used as proxy measures for improved care and quality of nursing homes. The program succeeded in reducing antipsychotic and anxiolytic prescribing by about 6 percentage points each from the approximately 33% rates for each before the PDSI was implemented.
A 6% absolute decrease, or an 18% relative decline, however, is only a modest reduction at best, and not the whole story. During this period, prescribing of anticonvulsants, antidepressants, and opioids increased substantially, by 12.4, 6.6, and 8.6 percentage points, respectively. Seemingly, over time, prescriptions for these medication classes practically replaced the decrease in use of antipsychotic and benzodiazepine prescriptions. The use of these other psychotropics increased such that by 2019, the end of the 10-year observation period, 63% of veterans with dementia in nursing homes were treated with antidepressants, 43% with anticonvulsants (mainly gabapentin), and 41% with opioids, while about 27% each were treated with antipsychotics and anxiolytics. The questionable or small changes in the prevalences of diagnosed depression, anxiety, and chronic pain do not appear to support the much larger increases in the prescribing of anticonvulsants, antidepressants, and opioids. Yet, more than two-third of veterans with dementia were also diagnosed with chronic pain, potentially justifying the use of drugs in these classes of psychotropics. An alternative to the idea of replacing antipsychotics with other psychotropics to treat agitation and psychosis is that physicians became more sensitive to the need to diagnose and treat chronic pain, and gabapentin, other anticonvulsants, and antidepressants were the vehicles for this.
There is irony in that a small shift away from antipsychotics that show modest evidence of clinical benefit in controlled trials resulted in the use of medications with minimal or absent empirical evidence of benefit. As the Michigan investigators noted, well-intentioned, administratively directed programs to simply reduce antipsychotics, rationalizing that this is indicative of better care, may lead to unintended consequences. The generally unmet need for effective treatments for behavioral and psychological symptoms of dementia is filled with other medications whose effectiveness and safety are uncertain. The concern about antipsychotics for people with dementia is with the drugs’ safety, cardiovascular adverse effects, and risk for death (
16), yet the medications that may be replacing them are less certain and have yet other safety risks.
Pimavanserin is an atypical antipsychotic with predominant 5-HT
2A antagonist or inverse agonist effects, a property it shares with quetiapine and clozapine, both used to treat psychosis in Parkinson’s disease, although the latter two have low D
2-blocking activity while pimavanserin does not. Pimavanserin has not demonstrated substantial evidence of effectiveness for agitation in dementia. In a randomized placebo-controlled trial for hallucinations and delusions in nursing home patients with Alzheimer’s dementia (
18,
19), pimavanserin showed questionably significant improvement in hallucinations at 6 weeks of treatment but not at 3, 9, or 12 weeks. A randomized placebo-controlled withdrawal trial in patients with hallucinations and delusions associated with dementia showed rapid relapse of symptoms after withdrawal, but the effect was in the subset of patients with Parkinson’s disease and not in those with Alzheimer’s or other dementias (
20).
Pimavanserin received FDA marketing approval in 2016 for hallucinations and delusions associated with Parkinson’s disease, based on one 6-week randomized placebo-controlled trial of 185 patients that was considered to be strongly positive, when ordinarily two positive trials would have been required by the FDA, evoking controversy both within and outside the FDA (
21). Pimavanserin’s label has the same boxed warning as other antipsychotics, citing an increased risk for death. It also carries a warning on increases in QT interval or prolonged ventricular repolarization, a property shared with many psychotropics and other drugs, which itself can lead to fatal cardiac arrhythmias. Since pimavanserin’s approval, concerns about its safety have been expressed by consumer watchdogs such as the Institute for Safe Medical Practices (
22) and the popular press (
23). An analysis of a Medicare long-term care database (
24) found an increased risk for death in patients with Parkinson’s disease treated with pimavanserin compared with no treatment over the course of 1 year.
Also in this issue of the
Journal, Mosholder and colleagues (
25), many from the FDA, compared Medicare beneficiaries who had Parkinson’s disease and received prescriptions for pimavanserin with those who received other atypical antipsychotics during the first 3 years that pimavanserin was marketed. Mortality was about 35% lower with pimavanserin during the first 6-month follow-up period than with atypical antipsychotics (78% received quetiapine). Beyond 6 months, however, pimavanserin no longer showed a mortality advantage, and nursing home patients experienced no advantage with pimavanserin, either short-term or longer-term.
Some caveats prevent concluding that pimavanserin is a safer alternative to (mainly) quetiapine for Parkinson’s disease psychosis. Patients who initiated alternative atypical antipsychotics were older; more likely to be treated by primary care specialists; more likely to have depression, anxiety, dementia, cardiovascular disease, diabetes, pulmonary disease, and renal disease; more likely to be receiving another drug that prolongs the QT interval; and more likely to have received prescriptions for antianxiety or antidepressant drugs. In short, they are of a group with a higher risk for death than those who were treated with pimavanserin. Notably, about 75% of patients did not continue either medication beyond 6 months, which suggests that for most patients any clinical benefit from pimavanserin or other atypical antipsychotics was time limited or absent. Considering that the only efficacy evidence for pimavanserin comes from a 6-week trial for Parkinson’s disease, and that there have been no completed postmarketing trials since it entered the market 6 years ago, evidence for the clinical benefit of pimavanserin is still in doubt. In the absence of clinical benefit, no drug is safe. These two reports illustrate the hazards and uncertainties of not carefully assessing both effectiveness and safety within the context in which drugs will be used.
As if to emphasize this, on June 17, 2022, an FDA advisory committee voted 9 to 3 that there was not substantial evidence for the effectiveness of pimavanserin for the treatment of hallucinations and delusions associated with Alzheimer’s disease, essentially recommending that pimavanserin not receive FDA approval for this purpose.