Abstract
Background: Guidelines, policies, and warnings have been applied to reduce the use of medications for behavioral and psychological symptoms of dementia (BPSD). Because of rare dangerous side effects, antipsychotics have been singled out in these efforts. However, antipsychotics are still prescribed “off label” to hundreds of thousands of seniors residing in nursing homes and communities. Our objective was to evaluate how and why primary-care physicians (PCPs) employ nonpharmacologic strategies and drugs for BPSD.
Methods: Semi-structured interviews analyzed via template, immersion and crystallization, and thematic development of 26 PCPs (16 family practice, 10 general internal medicine) in full time primary-care practice for at least 3 years in Northwestern Virginia.
Results: PCPs described 4 major themes regarding BPSD management: (1) nonpharmacologic methods have substantial barriers; (2) medication use is not constrained by those barriers and is perceived as easy, efficacious, reasonably safe, and appropriate; (3) pharmacologic policies decrease the use of targeted medications, including antipsychotics, but also have unintended consequences such as increased use of alternative risky medications; and (4) PCPs need practical evidence-based guidelines for all aspects of BPSD management.
Conclusions: PCPs continue to prescribe medications because they meet patient-oriented goals and because PCPs perceive drugs, including antipsychotics and their alternatives, to be more effective and less dangerous than evidence suggests. To optimally treat BPSD, PCPs need supportive verified prescribing guidelines and access to nonpharmacologic modalities that are as affordable, available, and efficacious as drugs; these require and deserve significant additional research and payer support. Community PCPs should be included in BPSD policy and guideline development.
Dementia is diagnosed based on cognitive decline with evidence of memory loss, but noncognitive psychiatric symptoms are seen in dementia of all etiologies and stages.1,2 Referred to as behavioral and psychological symptoms of dementia (BPSD), these include disordered mood, psychosis, inappropriate behaviors, and motor symptoms.1,3 These noncognitive neuropsychiatric symptoms are often the most intrusive, debilitating, and detrimental for patients and caregivers in terms of quality of life and health outcomes. Furthermore, BPSD symptoms are frequently critical and direct drivers of institutionalization and the economic cost of dementia.4,5
Although evidence proving the effectiveness of nonpharmacologic approaches for BPSD across domains of care is lacking, a growing body of literature demonstrates that these skills, techniques, activities, therapies, and personalized interventions can improve quality of life and help to manage certain symptoms and behaviors.6⇓–8 While these methods are safer for patients than medications, their individualized nature requires greater investments of time, effort, expertise, and financial resources.10 Also, reimbursement for these services is often poor. For all these reasons, they are likely underused in all settings including long-stay facilities.6,7,10⇓–12 These techniques are also more effective for some BPSD than for others.7 For example, an agitated patient can often be calmed by music, touch, and aromatherapy, especially in combination with other personalized approaches, whereas those who are hallucinating are more likely to need medication as part of the treatment plan.7,9,13⇓–15 At times even optimally applied nonpharmacologic therapies simply do not work because of numerous factors, including disease variation and progression.8,16
Medications used “off label” for BPSD include anxiolytics, antidepressants, cognitive enhancers, antipsychotics, and anticonvulsant mood stabilizers. Benzodiazepines, prescribed for anxiety and agitation, increase risks of sedation, confusion, falls, and fractures.17,18 Cognitive enhancers (N-methyl-D-aspartate antagonists and cholinesterase inhibitors) have been shown to have small statistically significant effects in ameliorating BPSD, but many patients already take these drugs.19,20 Serotonergic drugs can improve depression and anxiety but have limited efficacy for BPSD.21 Although they are generally well tolerated, even these medications convey potential risks; citalopram has been shown to be helpful for certain BPSD, but it blunts cognition and results in worrisome electrocardiographic changes.22 Anticonvulsant mood stabilizers are used for agitation and aggression in particular, although trials evaluating their efficacy and safety for frail elderly are scant, and results are mixed.23⇓⇓⇓⇓⇓⇓⇓⇓⇓⇓⇓⇓⇓–37 Some anticonvulsants, such as valproate and carbamazepine, carry US Food and Drug Administration (FDA) boxed warnings because of hepatic and bone marrow toxicities.38,39
Antipsychotic medications have the best evidence for improving BPSD, although they work only 25% to 30% of the time.40,41 While their benefit in BPSD is modest, their risk profile is extensive. In addition to extrapyramidal symptoms and sedation, antipsychotics are associated with confusion, falls, and aspiration events.42 However, increases in strokes, heart attacks, and mortality associated with the use of typical and atypical antipsychotics earned both FDA boxed warnings.42⇓–44
In addition to the FDA warnings, political and regulatory spotlights have focused on medication use for BPSD. Inappropriate antipsychotic prescribing for BPSD has been labeled “chemical restraint” and “elder abuse.”45 Some states have proposed legislation to require written informed consent for their use.46
The long-standing agenda of the Centers for Medicare & Medicaid Services (CMS) to curb drugs for BPSD was codified in the 1987 Nursing Home Reform Act, which prohibited the chemical restraint of patients in nursing homes.47 In March 2012, as part of a concerted effort to check off-label prescribing of antipsychotics for BPSD, CMS launched the National Initiative to Improve Behavioral Health & Reduce the Use of Antipsychotic Medications in Nursing Home Residents48, subsequently renamed the National Partnership to Improve Dementia Care in Nursing Homes. The rate of antipsychotic use in long-stay facilities was made a quality measure, published through the Medicare Nursing Home Compare website (https://www.medicare.gov/nursinghomecompare/search.html?), and factored into Medicare's 5-star facility rating. In nursing homes, pharmacists also supervise mandatory periodic reviews of all psychoactive medications to encourage consideration for weaning or reducing doses.48 Since the initiative's debut, antipsychotic use in nursing homes has decreased steadily. CMS's initial goal of a 15% nationwide reduction was achieved in the final quarter of 2013, and by the fourth quarter of 2016 the national rate was down 33%, from 23.8% to 16%.49,50
National and international guidelines for medical management of BPSD all state that antipsychotics should be used only when other measures fail and with appropriate discussion of the risks and benefits. According to the American Geriatric Society and respective national and provincial Canadian groups, only potential harm to self or others justifies their use.51,52 However, the American Psychiatric Association, the American Alzheimer's Association, and the British National Institute for Health Care and Excellence guidelines liberalize indications to include refractive patient distress.53⇓⇓–56
Despite evidence of only modest benefit and proven severe risk, the context of purposeful efforts from the FDA, CMS, and state legislation, and the censure of geriatric societies and patient advocacy groups, physicians continue to prescribe antipsychotics and other medications for BPSD to hundreds of thousands of patients in US nursing homes and communities.50 The reasons for this apparent conundrum are not well understood, and investigation in this area has been scant. This study sought to explore American primary care physicians' use of nonpharmacologic modalities and drugs to treat these challenging symptoms.
Methods
We adapted the qualitative research framework posited by Leeman and Sandelowski57 for “Designing Qualitative Studies to Elicit Practice-Based Evidence,” in this study examining “What aspects of the practice context may interact with interventions or strategies to moderate patient or system outcomes?”
We administered semistructured interviews of 26 participants (primary care physicians [PCPs]) to determine their experiences with and perceptions of medication and nonpharmacologic management of BPSD. (See the Appendix for an abbreviated interview guide.) Telephone interviews (1-on-1) averaged 45 minutes (range, 20–65 minutes). Participants were interviewed by family physicians (JWK, JDW) from their district (see ‘Setting’ below).
Participants and Setting
Our participants were 26 PCPs from the 5 counties of the Lord Fairfax Health District of northwestern Virginia who were involved in direct patient care >20 hours/week and had been in primary care practice within the district >3 years. We did not stipulate the percentage of patients with dementia included in the PCPs' patient panels. Participants were recruited in person, over the telephone, and through E-mail. An honorarium of $50 was offered. Our initial goal was to have 8 PCPs who cared for patients in nursing homes and 16 who did not, in order to reach data saturation within each group, and to have a diverse group of physicians with overall project saturation of around 24 participants.58,59 However, half our participants (13) had at least 1 patient in a nursing home facility. Of those 13, 10 had >5 patients in nursing homes, and these 10 were used for comparative analysis with the remaining 16. We did achieve data saturation within those groups and for the overall group in our final analysis.58 We further discovered that views of all PCPs were largely congruent, despite training in family or internal medicine, despite caring for patients in nursing homes (or not), and despite duration of practice. Results presented here (except where noted) reflect the entire PCP group.
Data Analysis
Transcriptions of audio-recorded interviews were organized and coded using Atlas.ti. Codes for use in template analysis were developed both inductively and deductively, yielding a priori codes that were honed and augmented by emergent findings.60 The coding team (JWK, KMW, TB) developed a preliminary code book based on the research questions and relevant literature. After 6 to 8 interviews, the team applied an immersion and crystallization process of reading and rereading transcripts to refine codes while also allowing previously unpredicted findings to be added to the code book.58 This process occurred iteratively until the code book was well defined and coding saturation was reached, in this case 3 cycles. Coding then proceeded independently for all transcripts, after which the team came together to resolve coding differences via negotiated consensus to ensure strong intercoder agreement.61
Preliminary themes were identified through the analysis of coding elements. Two authors (JWK, JDW) used a method of metaphor and allegory described by Ryan and Bernard62 to further delineate and appraise code patterns and prospective significant “themes and subthemes”: “Analysis … becomes the search for metaphors in rhetoric and deducing the schemas or underlying themes that might produce those metaphors.” These were inspected iteratively over 3 cycles of meetings, during which disconfirming evidence was sought for those themes and differences were resolved through negotiated consensus. From these metaphors 4 major themes emerged and were organized into the findings presented in the Results. All authors vouch for the validity of these findings.
Dr. Etz audited the ongoing analytic process and guided the team's efforts. This study was approved by the Institutional Review Board of Virginia Commonwealth University, Richmond.
Results
Participants
We interviewed 26 PCPs, 16 trained in family medicine and 10 in internal medicine. Among the internists, 2 had additional qualifications in geriatrics. Our participants included 19 men and 7 women, with an mean age of 50 years (range, 35–69 years) and an average practice duration of 19 years (range, 4–38 years). They self-identified as white (n = 21), Asian (n = 3), and African American (n = 1); 1 participant declined to comment. Three had 1 patient in a nursing home and 10 had >5 patients in a nursing home (mean, 18 patients; range, 6–30 patients). Of the 10 PCPs with >5 patients in a nursing home, 6 were family physicians.
Themes
Four major themes were identified. Exemplar quotes are presented in Table 1.
Primary Care Physician Exemplar Quotations, by Theme
Theme 1: Nonpharmacologic BPSD Therapies Have Substantial Barriers
All PCPs described services supporting dementia care as inadequate. That said, PCPs reported that few patients fully capitalized on all available resources. Family capabilities and involvement were noted as significant factors affecting optimal utilization. While all families were characterized as increasingly overwhelmed as the disease progressed, some families were depicted as extremely proactive and much more successful in identifying and using national and local resources. However, a majority of families were reported to struggle to identify and use these existing supports.
Most PCPs acknowledged they had little formal training in nonmedication therapies for dementia symptoms and that time constraints made any personalized instruction by them untenable. In addition, the changing landscape of community assistance and resources was described as so variable that maintaining awareness of current availability was nearly impossible. The few services that PCPs could describe as consistent, available, and helpful (for example, “in-home services” from a geriatric organization) were characterized as overwhelmed by demand.
Similar to the services currently available to patients with diabetes via trained diabetes educators, some PCPs imagined a “navigator” or counselor who could serve as a guide to shepherd families through BPSD or for all aspects of dealing with the disease process. All nonpharmacologic measures (eg, hired caregivers, assisted living) were acknowledged to involve significant, often prohibitive, costs.
PCPs stated that nonpharmacologic therapies are essential first-line measures that are helpful for BPSD to some extent all the time. However, PCPs viewed these modalities as having only about a 50% efficacy to relieve bothersome symptoms and as eventually losing effectiveness as dementia and BPSD worsened. The PCPs also lamented that the evidence for the efficacy of nonpharmacologic BPSD therapies lacked the robustness and generalizability they desired.
Theme 2: Medications Are “Easy,” Effective, Safe, and Address Patients' Needs
In stark contrast to nonpharmacologic modalities, no PCP reported cost as a barrier to medication use. Medications were said to be the easiest BPSD solution for patients, caretakers, and doctors: they are affordable, easily available, require minimal training, and quick to take effect.
In addition to being “easy,” almost all PCPs indicated that medications worked at least half of the time, even for recalcitrant symptoms. Severe side effects were almost never knowingly discerned, and all classes of medications seemed to convey roughly the same collection of adverse effects, of which the most concerning and common were sedation and falls. Specifically, of the 26 PCPs interviewed, only 2 described direct experience with possible severe side effects from a medication. One PCP from the nursing home group described a colleague's patient who experienced a stroke soon after initiating antipsychotic therapy. Because stroke is not rare among patients in nursing homes, he wondered, but was uncertain, whether the medication was the cause. Another nursing home physician described a patient receiving a large dose of quetiapine and who had a seizure soon after arriving to his dementia unit. Because seizures occur with some regularity in this population, he suspected, but again was unsure, that the medication played a role. Despite these experiences, both physicians continued to prescribe antipsychotic medications for BPSD. Otherwise, PCPs reported not seeing the dire effects of antipsychotics and other medications prescribed for BPSD described in FDA black box warnings. Except for occasional episodes of extrapyramidal symptoms, antipsychotics were said to have the same side effect and safety profiles as other medications used for BPSD.
PCPs stated that medication use was both crucial and appropriate to address 4 areas of patient needs: distress, safety, functionality, and care maintenance. Symptoms of distress often overlapped with safety, including agitation, paranoia, belligerence, hostility, aggressiveness, and wandering. Functionality was another important goal that PCPs viewed as justifying pharmacologic therapies. This included assisting sleep and allowing for interactions with others. Most PCPs asserted that using medications for BPSD allowed patients to remain at home or in assisted living, delaying unwanted transitions to higher levels of care.
Theme 3: Consequences of Current Policies
Physicians in practice >10 years acknowledged that they used medications for BPSD, including antipsychotics and benzodiazepines, less often and in lower doses than they previously had. Physicians with less experience did not note such changes. Most PCPs working in nursing homes stated that they are regularly challenged by required pharmacy queries about the use of all “psychiatric” or “behavioral” medications. These reviews, along with an increased awareness of risk from information published in journal articles and continuing medical education courses, were said to be among the reasons that they changed both their documentation and their prescribing. When using drugs for BPSD, almost all PCPs reported currently using more scrupulous documentation that details any discussion with families as well as why the medication was required. Most nursing home PCPs even adjusted their diagnoses to better support medication use, for example, “dementia with psychosis” when prescribing an antipsychotic. PCPs reported trying to minimize antipsychotic use for BPSD. However, they also reported increasing their use of other medications that had rarely been used for dementia symptoms, particularly singling out sedating antidepressants such as trazadone and anticonvulsant mood stabilizers such as valproate as examples of such changes. PCPs noted that psychiatric and neurologic consultations both in the outpatient setting and in nursing homes helped relieve diagnostic and administrative time burdens while shifting the risk of off-label prescribing to specialists. However, PCPs agreed that consultants used medications for BPSD as much or more than they would have.
PCPs described discussions with families about treating BPSD with medications as nuanced and challenging. Most PCPs acknowledged that this conversation was often longer and more detailed than in the past, often including an analysis of risks and benefits. However, PCPs found that such discussions rarely resulted in medications not being used. PCPs stated that most community caregivers came to them with an urgent request for help only when they had exhausted known services and nonpharmacologic options. For long-stay patients, PCPs described significant variation between requests for medications from different facilities, and they were, in general, were more likely to expedite medication requests from trusted personnel whom they knew had already checked for other potential causes of BPSD and had applied maximal nondrug efforts.
Theme 4: PCPs Need Supportive Evidence-Based Guidelines
Many PCPs commented on the lack of proven guidelines for nonpharmacologic interventions for BPSD. Furthermore, almost all PCPs expressed frustration about not being adequately supported by evidence or assisted by guidelines and policies with regard to how to approach severely intrusive but not life-threatening BPSD when nonpharmacologic measures fail. All PCPs believed that medications are sometimes appropriate to treat patient-oriented palliative goals, to relieve distress, or to improve quality of life, even in the absence of a direct threat of harm. All voiced dissatisfaction with guidelines or policies that, rather than assisting them in meeting such patient needs, characterized them as “bad doctors.”
Discussion
PCPs identified many practical and logistic barriers to accessing nonpharmacologic therapies for their patients, of which money, time, and convenience were most significant. In direct contrast, no significant barriers were described to prescribing drugs for BPSD. Clearly, medication use is not likely to decline until alternative therapies receive better support.
Inherent in the PCP description of barriers to nondrug therapies is the conviction that medication use for BPSD can be reduced in all care settings through optimal use of these alternative modalities. Although this would necessitate overcoming the obstacles enumerated above, PCPs identified an opportunity to enhance such care simply by improving the use of services that are already available. That few families take full advantage of fragmented existing resources is consistent with what community caretakers have previously described.16 In addition, congruent with this suggestion, dementia “navigators” in Minnesota and “counselors” elsewhere have been shown to be of great benefit to families, optimizing support, personalizing therapies, and improving both patient and caregiver outcomes.63
When alternative approaches failed or were unavailable, almost all PCPs readily used drugs, including antipsychotics, to address BPSD. Similar to other physician studies, but contrary to the pharmacologic literature, PCPs felt that medications worked the majority of the time.37,64 Contrary to data regarding harms, PCPs viewed medications as generally safe. No particular medication class was observed to have a singular risk profile, and no PCP was certain they had witnessed a “black box warning” event. As previously proposed, it may be that it is difficult to perceive rare medication-related effects against the backdrop of severe chronic morbidity.16 For example, distinguishing a 12-week survivorship rate of 97.7% from 96.5% for patients taking antipsychotics may not be clinically appreciable.65
The discrepancy between PCPs' perceptions about medication efficacy and safety compared with what is proven may contribute to persistent medication use. How much increasing education about medication efficacy and side effects will change prescribing is unclear, because most PCPs showed reasonably detailed awareness of risks and warnings. This phenomenon is not unique to American PCPs; Dutch geriatricians, who clearly acknowledged expected increased morbidity and mortality from antipsychotic use, continued to prescribe them for “agitation” and “aggression.”64 Similarly, increasing availability of specialty consultants for BPSD management may not reduce prescribing, because although PCPs viewed specialists as helpful, they described them as using antipsychotics and other medications at least as often as the referring PCPs.
In order to prescribe antipsychotics for someone with dementia, informed decisions are touted and legislation requiring informed consent has been proposed.46 However, the impact this would have on management approaches is unknown. Many PCPs in this study acknowledged having a different conversation now than a decade ago with families about treatment options for BPSD, often including more detailed discussion of benefits and risks. It is interesting, though, that PCPs reported that even detailed discussion of risks and adverse effects rarely resulted in a decision not to use medication. In addition, the amount of autonomy caregivers want in medication decisions varies from actively evaluating pros and cons and being the primary decision maker to deferring all medication decisions to the physician.16,65
FDA warnings, CMS efforts, and guidelines about antipsychotic use in patients with dementia have had multifaceted consequences for almost all the PCPs we interviewed. In nursing homes, periodic medication reviews overseen by pharmacists seemed to be effective in encouraging dose reductions and weaning. One resultant change was a decrease in antipsychotic use. Other consequences included changes in charting, changes in how neuropsychiatric diagnoses are reported, and an increase in the use of alternative drugs for BPSD which have real but not fully evaluated dangers. These unintended consequences merit more detailed evaluation and should be considered for future research.
PCPs all expressed frustration with policies and guidelines that did not support them in the management of intrusive symptoms when nonpharmacologic therapies fail. All PCPs felt that circumstances exist where medications are appropriate even in the absence of threat of harm. Beyond symptoms, PCPs felt that quality of life and other patient-oriented goals have to be considered when prescribing. A previously underappreciated patient-oriented goal was the desire for the patient to stay in their own home. This is congruent with caregiver stakeholders who also had a palliative perspective for dementia care.16 While some policies and guidelines would endorse all such prescribing, it is contrary to others, and PCPs resent being characterized as “bad doctors.”45 Ongoing and appropriate medication reduction efforts may need to take these findings into consideration.
All PCPs expressed a need and desire for practical and evidence-based guidelines for how to effectively help patients with BPSD, both in their homes and in facilities, and including both medications and nonpharmacologic approaches.
Strengths and Limitations
Our study's strengths include a diverse sampling of experienced PCPs in the Lord Fairfax Health District of Virginia, who represent family and general internal medicine and who do and do not work with a significant number of patients in nursing homes.
Weaknesses include that the district is rural and our physicians were older, and mostly white men. Lacking are the perspectives of more ethnically diverse urban and suburban PCPs as well as other providers who actively care for patients with dementia. Teasing out differences between physicians in family medicine, physicians in internal medicine, and other practitioners, as well as variations in approaches used in different settings, may be feasible with larger investigations in other locales.
Conclusions
PCPs continue to prescribe medications for symptoms of dementia because they view them as safer and more effective than controlled studies report. They use drugs for symptoms with a direct threat of harm but also to meet patient-oriented goals, including easing patient suffering. PCPs endorse that increasing nonmedication methods for treating BPSD would decrease the need for drugs, but they also describe many barriers to the use of these methods. To decrease medication use, nonpharmacologic therapies will need to be as easy, available, and affordable as drugs—but this requires research and funding. Opportunities may exist to optimize the use of resources already present in communities. Increasing education about medication risks and increasing specialty support were not identified as factors likely to reduce medication use. Those who design guidelines and policies should consider recruiting community PCPs to participate in the development of clinically meaningful recommendations.
Acknowledgments
The authors thank the Shenandoah Valley Family Practice Residency for additional support. The authors are grateful to the family and internal medicine physicians of northwestern Virginia for their participation in this project, and to Christine Carlson Kerns for her assistance with early analysis. We are also grateful to the VCU Department of Family Medicine and Population Health, and to Dr. Anton Kuzel and Dr. Rebecca Aycock for their assistance.
Notes
This article was externally peer reviewed.
Funding: This study was funded in part by Virginia Commonwealth University's Clinical and Translational Science Award (UL1TR000058) from the National Center for Advancing Translational Sciences and the Center for Clinical and Translational Research's Endowment Fund of Virginia Commonwealth University.
Conflict of interest: none declared.
To see this article online, please go to: http://jabfm.org/content/31/1/9.full.
- Received for publication June 14, 2017.
- Revision received August 18, 2017.
- Accepted for publication August 23, 2017.
