Readers of this journal are undoubtedly familiar with the challenges associated with managing the behavioral and psychological symptoms of dementia (BPSD). But, to briefly summarize: Although cognitive impairment is the clinical hallmark of dementia, BPSD such as apathy, delusions, agitation, and sleep disturbances are exceedingly common, occur in all forms of dementia, and often dominate disease presentation (Lyketsos et al., Reference Lyketsos2011). Nearly all patients will exhibit these symptoms at some point in the course of dementia (Geda et al., Reference Geda2008; Mega et al., Reference Mega, Cummings, Fiorello and Gornbein1996). Such symptoms, as opposed to core cognitive symptoms, are particularly vexing and create the most difficulties for patients, caregivers, and clinicians, leading to earlier nursing home placement (Stern et al., Reference Stern1997; Yaffe et al., Reference Yaffe2002). Just one or two BPSD symptoms are associated with a doubling (an additional 10 hours weekly) of active help received from caregivers (Okura and Langa, Reference Okura and Langa2011).

While non-pharmacological interventions are intended to be the first-line treatments for BPSD, it is unclear how much this actually happens in practice (Kales et al., Reference Kales, Gitlin and Lyketsos2015). Instead, persons with dementia are awash in psychotropic medications. For example, in a recent analysis of a nationally representative sample in the USA, the percentage of adults with dementia prescribed an antidepressant ranged from 34.3% in the community to 64.5% in nursing homes (Lei et al., Reference Lei, Samus, Thomas and Maust2022), prescribing that far exceeds that to the general population of older adults (Moore and Mattison, Reference Moore and Mattison2017). In the community, 19.9% of adults living with dementia were prescribed a benzodiazepine and 12.0% an antipsychotic, 32.6% and 28.0%, respectively, in nursing homes—again, far higher than prescribing to those without dementia. In addition to monotherapy, psychotropic polypharmacy is worryingly high in both community and long-term care settings (Jester et al., Reference Jester, Molinari, Zgibor and Volicer2021; Maust et al., Reference Maust2021). While some of these medications may be prescribed for off-label indications other than dementia (e.g. gabapentin for neuropathic pain) or pre-existing psychiatric illness, some amount of the psychotropic use is undoubtedly an attempt to treat BPSD.

Unfortunately, clinicians considering pharmacological treatments for agitation in their patients with dementia have relatively few options, and antipsychotics, the class with the most evidence supporting efficacy, also have the most significant associated risks (e.g., mortality). The array of agents that are being prescribed may partially reflect clinicians’ attempts to avoid antipsychotics. In a survey of clinicians that specifically queried antipsychotic substitutes, the most-endorsed potential substitute was valproic acid, despite multiple negative trials and a Cochrane review recommending against its use for agitation (Baillon et al., Reference Baillon, Narayana, Luxenberg and Clifton2018; Olivieri-Mui et al., Reference Olivieri-Mui, Devlin, Ochoa, Schenck and Briesacher2018). Indeed, analyses of prescribing in long-term care settings in both the USA and Canada have demonstrated that, while antipsychotic use declined, use of antiepileptics has increased (Harris et al., Reference Harris2022; Maust et al., Reference Maust, Kim, Chiang and Kales2018). However, surveyed clinicians also identified a host of other potential substitutes, including various antidepressants, benzodiazepines, prazosin, and buspirone. Clinicians appear to be resorting to a grab-bag of psychopharmacology in the interest of just doing something.

One key advance on this front has been the development of pharmacological treatment algorithms for agitation or aggression in dementia, such as that from Davies and colleagues (Davies et al., Reference Davies2018). The key initial first step of this algorithm is stopping medications that were specifically started for BPSD. Therapeutic inertia is a powerful force, and clinicians (and family members) may be reluctant to stop ongoing therapies for fear of further worsening symptoms. However, given the relatively limited evidence base for the benefit of any agent for BPSD weighed against the extensive evidence of harms including falls and further cognitive impairment, medications without clear evidence of benefit should be stopped. Then, starting from a relatively blank slate, clinicians start with antipsychotics rather than a smorgasbord of psychopharmacology with the primary virtue of not being an antipsychotic. The algorithm proceeds through a sequence of subsequent prescribing options, with decision points informed by measurement-based care.

In this issue of the journal, Kumar et al. (Reference Kumar2022) have built on this algorithm and implemented it as part of an integrated care pathway (ICP) for patients with Alzheimer’s or mixed dementia admitted to an inpatient geriatric psychiatry unit for management of agitation or aggression. As enacted by their team, patients first received a thorough medical and psychiatric workup to address potential underlying conditions. This was followed by individualized non-pharmacological interventions at the same time as ineffective medications were being tapered. Finally, for patients who were still symptomatic, they followed the steps of the Davies medication algorithm. The analysis considered symptom burden (e.g., the Cohen-Mansfield Agitation Inventory and Neuropsychiatric Inventory Questionnaire) along with length of stay, psychotropic polypharmacy, and documented falls. They included three different control groups, the first of which was a historical control of patients with dementia and agitation admitted before the ICP protocol was initiated. To consider potential secular trends in treatment in the inpatient geriatric psychiatry setting, the authors considered a change in the outcomes of interest using patients with a mood disorder (but not dementia) and agitation from the same time periods as the ICP and historical control dementia groups.

In a relatively small pilot study, it would be wrong to make conclusions about the efficacy of the ICP intervention, though the results are certainly promising. The level of agitation in ICP patients declined significantly, an improvement achieved while also reducing polypharmacy at the time of discharge relative to their historical controls (i.e. adjusted odds ratio 0.17), albeit with a large confidence interval secondary to the small sample size. As might be expected with such a reduction in polypharmacy, the ICP patients had a significantly lower risk of experiencing a fall during their admission relative to their historical controls. And, based on the analysis of the patients with mood disorders, these changes did not simply reflect secular trends that occurred in this setting over this time.

A paper presenting a pilot study typically concludes with a call for a larger, multisite study to definitively establish the efficacy of the intervention. Indeed, this is how Kumar and colleagues end their paper—and hopefully, the authors have the opportunity to do just that. However, do we really need to wait years for funders to approve and then investigators to complete what would undoubtedly be a complex, multisite study? Even while we wait for such a definitive trial, it is unclear why this algorithm or a similar methodical, standardized approach should not simply be the standard of care on inpatient settings caring for patients with dementia and agitation.