Implementation and Effectiveness of Cognitive Behavioral Therapy for Insomnia in Geriatric Primary Care

ABSTRACT Objectives We evaluated a plan for implementation and effectiveness of cognitive behavioral therapy for insomnia (CBT-I) in geriatric primary care by a geropsychologist. Methods The flow of referrals to a geropsychologist was tracked and, among those eligible and interested in participating, success in deprescribing sleep medications and the effectiveness of CBT-I were documented. Results Seventy patients were referred for evaluation of whom 62 were eligible for CBT-I; 34 began CBT-I and 29 completed a full course of treatment. Almost two-thirds of treatment completers were the “old old” (76–84 years) and “oldest old” (85–93 years) with multiple medical problems. Most treatment completers taking sleep medications had them deprescribed at the beginning of treatment and, one year after treatment, did not have them re-prescribed. After CBT-I, two-thirds of patients met the insomnia severity index criteria for response; and three-fifths for remission from insomnia. Further, most patients had sustained improvement in their target insomnia symptom(s) and sleep efficiency. Conclusions CBT-I can be implemented in geriatric primary care with successful deprescribing of sleep medications and meaningful improvement in symptoms of insomnia in a group of older adults of advanced age with multiple medical problems. Clinical Implications Clinical gerontologists can play an important role in improving late life insomnia.


Introduction
Acute and chronic insomnia are common in older adults.Insomnia is characterized as problems with falling asleep, staying asleep, and/or waking earlier than intended about which there is significant distress or impairment (American Psychiatric Association, 2013).Recent reviews suggest that the prevalence of insomnia symptoms in older people ranges from 30-48% and that for insomnia disorders the prevalence is from 12-20% (Patel, Steinberg, & Patel, 2018).Good sleep is a critical element of brain health (Dzierzewski & Dautovich, 2018) and reduces falls in older adults (Stone, Ensrud, & Ancoli-Israel, 2008).Sedatives and hypnotics are often prescribed to older adults with insomnia despite longstanding concerns about the adverse effects of these medications including risk for falls (Bloom et al., 2009;Fick et al., 2019).Cognitive behavioral therapy for insomnia (CBT-I) has been found to be highly efficacious both for younger and older adults although the vast majority of efficacy studies of older people have been with the "young old" (ages 65-74;McCurry, Logsdon, Teri, & Vitiello, 2007;Rybarczyk et al. 2013;Trauer, Qian, Doyle, Rajaratnam, & Cunnington, 2015).Further, CBT-I has a more durable response than sleep medications (Mitchell, Gehrman, Perlis, & Umscheid, 2012).Despite the strong recommendation of professional organizations that CBT-I should be the first-line treatment for insomnia rather than sleep medication, the movement of CBT-I from clinical research studies to widespread implementation in practice, particularly in primary medical care, has been sorely disappointing (Brasure et al., 2016;Qaseem, Kansagara, Forciea, Cooke, & Denberg, 2016).
Incorporation of many evidence-based treatments into primary care and other settings has been challenging.The field of implementation science has addressed these challenges by documenting the means by which evidence-based practices can be integrated into care settings and by evaluating their effectiveness in clinical populations (Bauer, Damschroder, Hagedorn, Smith, & Kilbourne, 2015).The chief impediments in access to and use of CBT-I in primary care appear to be "systemic," "clinician," and "patient" barriers (Koffel, Bramoweth, & Ulmer, 2018).Among the systemic barriers are few providers trained in CBT-I (probably less than 800 in the world; Thomas et al., 2016) and a limited number of venues for formally learning CBT-I.Clinician barriers are reflected in less than adequate screening for insomnia in primary care and physician uncertainty about the effectiveness of CBT-I.Patient barriers include unwillingness to engage in CBT-I because of effort required to engage in it (vs apparent relative ease of taking a sleep medication; Koffel et al., 2018).
In a previous brief report, we described pre-post treatment group differences in CBT-I outcomes which were generally large and meaningful among our older patients (Hinrichsen & Leipzig, 2021).These included improvement in the insomnia severity index (ISI; (Morin, Belleville, Belanger, & Ivers, 2011), sleep onset latency (SOL), wake after sleep onset (WASO), early morning awakening (EMA), and sleep efficiency (SE), sleepiness, as well as anxiety and depression symptoms.However, group level differences may lack specificity in discerning clinically meaningful changes in individuals -notably individual patient treatment response and treatment remission.This paper reports on an effort to implement CBT-I into geriatric primary care.We used existing clinical data to describe aspects of the implementation plan, success in deprescribing sleep medications, and the effectiveness of CBT-I based on ISI measured treatment response and relapse.We share observations about how we attempted to address systemic, clinician, and patient barriers to implementation.

Design
The Brookdale Department of Geriatrics and Palliative Medicine is an academic department within the Icahn School of Medicine at Mount Sinai in New York City.Among its missions is training physicians, conducting research, and providing clinical services in a variety of programs.In provision of clinical services to older adults with insomnia, our program was cognizant that AGS Beers Criteria for potentially inappropriate medication for older adults include sedatives and hypnotics (Fick et al., 2019).The practical challenge in our practices was that behavioral treatments for insomnia were not easily accessible within or outside of our department.A decision was made to build that expertise within our department and then deliver behavioral insomnia treatment services in one of our clinical programs.Success of that implementation plan depended on access to high quality training in an evidence-based insomnia treatment for a staff member, engagement of healthcare providers to make insomnia referrals, patient willingness to take part in a behavioral treatment for insomnia and demonstration of effectiveness of the treatment.

Intervention
The department geropsychologist (the first author), received formal training in CBT-I through the behavioral sleep medicine program, University of Pennsylvania, and delivered care guided by the treatment manual, Cognitive Behavioral Treatment of Insomnia: A Session-by-session Guide (Perlis, Jungquist, Smith, & Posner, 2008).The three core components of this eight session protocol are: sleep restriction (restricting the opportunity to sleep to the average nightly hours the patient was sleeping prior to starting CBT-I), stimulus control (using bed only for sleep and sex), and sleep hygiene (following basic principles to promote better sleep such as being mindful of the use and timing of caffeine and alcohol, establishing a regular sleep time, reducing noise and extraneous light).Cognitive therapy (e.g., challenging anxiety provoking thoughts about sleep) may be part of treatment but is considered by Perlis et al. as a "second line intervention."For over a year, the geropsychologist received formal consultation on CBT-I insomnia cases from a nationally recognized insomnia specialist to build and sustain adherence to the CBT-I protocol.Training and consultation were paid by departmental funds and some support from a foundation grant for the treatment of depression.

Participants
The medical primary care practice from which older patients were referred during the years 2015-2018 is the Martha Stewart Center for Living (MSCL).The MSCL is the outpatient geriatric primary care practice of Mount Sinai Hospital.The MSCL employs a range of healthcare professionals and has 4,500 enrolled patients whose median age is 85 years.Prior to the beginning of this initiative, the geropsychologist already was providing the equivalent of one day of clinical service in the MSCL.Referred patients were evaluated by the geropsychologist in an initial diagnostic and assessment intake to judge whether they were appropriate for CBT-I.Those with untreated obstructive sleep apnea or who did not meet criteria for insomnia did not receive CBT-I.CBT-I was offered to eligible older patients and, if interested, they began a course of CBT-I.Patients taking sleep medications who were interested in discontinuing them had their medications tapered by the prescribing geriatrician prior to beginning CBT-I.Participants had weekly, in-person CBT-I sessions.Services were eligible for reimbursement by Medicare.

Measures
Clinical data formed the basis for tracking patient referrals, engagement and compliance with CBT-I, judging success of deprescribing sleep medications, and documenting the effectiveness of CBT-I.This study of existing clinical data was approved by the Icahn School of Medicine at Mount Sinai IRB.Clinical record derived data discussed in this report include demographic, medical, mental health, and sleep-related indicators.
Demographic variables included patient age, sex, race, education level, and number of listed medical problems and medications in the electronic medical record (EMR).Duration of insomnia and current use of medications for sleep were ascertained during the initial intake.The geropsychologist determined during intake whether patients currently met criteria for a DSM-5 diagnosis of insomnia disorder and any other mental disorder.Older patients who participated in CBT-I completed a weekly standardized sleep diary that contained the essential elements of a "the consensus sleep diary" (Carney et al., 2012) throughout the course of treatment.The content of the diary was entered into an Excel spreadsheet created by James Findley and Michael Grandner which provided summary values for key insomnia-relevant parameters.
At the start of treatment, each older adult identified whether the focus of the treatment would be SOL (minutes to fall asleep), WASO (minutes awake after falling asleep), EMA (minutes awake earlier than intended), or a combination of these.Also, the ISI was administered at the start of treatment.The ISI consists of seven questions each scored from zero to four, resulting in a range of 0-28, with 0-7 considered no clinical insomnia, 8-14 subthreshold insomnia, 15-21 moderately severe insomnia, and 22-28 severe insomnia (Morin et al., 2009).Patients were considered "treatment responders" if the pre-post change in ISI was greater than 7; and patients were considered "treatment remitters" if, at end of treatment, the ISI score was less than 8.Other outcomes reported here include the percentage of CBT-I completers who achieved an average 50% reduction or more in their respective individual or combined target symptoms for the final two weeks of treatment as derived from their sleep diaries, and whether the average of the final two measures of sleep efficiency (SE; the percentage time in bed spent sleeping; range 0-100%,) was 85% or better.
We documented the percentage of patients who discontinued their sleep medications prior to the start of CBT-I as well as the percentage of these who, one year after completing CBT-I treatment, did not have a prescription for a sleep medications noted in the EMR.Descriptive statistics (frequency counts and measures of central tendency), t-tests and chi-squares were used to analyze the data.

Engagement of patients
Seventy patients completed the initial diagnostic and assessment intake.As indicated in Figure 1, eight were excluded because they did not meet DSM criteria for insomnia or other reasons noted in the figure (American Psychiatric Association, 2013), and 28 patients decided not to participate in CBT-I for the reasons given in Figure 1 ("nonparticipants").There were 34 participants (55% of referrals).Of these, 29 (85%) completed the therapy and five dropped out because they found the CBT-I protocol too difficult to do.Session 1 was the intake which was followed by 1-3 sessions to review baseline sleep data.Typically, more baseline sessions were needed for patients who were tapering off sleep medications.After the intake and these baseline sessions, the mean number of CBT-I sessions for completers was 6.1 (SD = 3.3; range 2-16 weeks; median = 5).Most end of treatment data were available for 28 patients and information about possible continued use of sleep medications for 13 of the 14 patients who were still in the practice one year later.

Engagement of geriatricians to make referrals
A potential clinician barrier to implementation of CBT-I may be their screening of and referring patients for insomnia.The geropsychologist initially met with clinic staff to provide education about CBT-I as a treatment modality and establish a system for making referrals for it.Geriatricians made referrals by the EMR.Among those patients discontinuing sleep medications, geriatricians guided the patient in tapering those medications.The geropsychologist periodically met with clinic staff to discuss any patient issues and gave some formal presentations on CBT-I to members of the department to further build knowledge and enthusiasm about this treatment.

Deprescribing of sleep medications
As can been seen in Table 1, 55.2% of treatment completers were taking sleep medications prior to commencing CBT-I.Eighty-seven percent of those taking sleep medication at baseline were able to discontinue the medication before starting CBT-I based on a deprescribing regimen provided by their respective geriatricians.One year after completion of CBT-I, 69.2% had no prescription for a sleeping medication in their respective EMRs.

Effectiveness of CBT-I
As summarized in Table 1, two-thirds of patients were "treatment responders" and 62.9% were "treatment remitters" based on the ISI.As noted, similar numbers of completers identified each of the three target symptoms SOL, WASO, and EMA, as the focus of their treatment; and most had more than one target symptom Over the last two weeks of treatment, 50% reduction or more in minutes was achieved by 68% of completers who targeted SOL, 77% who targeted WASO, and 81% who targeted EMA.Among the 20 patients who had more than one target symptom, 55% achieved a 50% reduction or greater in minutes in all targeted symptoms.Sleep efficiency of 85% or greater was achieved in 71% of completers over the last two weeks of treatment.These outcomes did not significantly differ between for the "young old" and the combined group of "old old" and "oldest old" completers based on chi-squared analyses

Overcoming barriers to implementation of CBT-I
Implementation of CBT-I in this geriatric primary care practice with a large group of "old old" and "oldest old" was generally successful.A potential systemic barrier was addressed by the provision of administrative support to the geropsychologist to obtain formal CBT-I training and case consultation.A steady flow of referrals from geriatricians for CBT-I and deprescribing of sleep medications were evidence of overcoming potential clinician barriers of screening and referral for treatment of insomnia.
However, an existing "patient" barrier to implementation was evident in the unwillingness of almost half of eligible older adults to actually start CBT-I.Although patients self-reported a variety of Total N = 29 but some variables do not total 29 because of missing data or indicator is a subset of the total sample.a This variable has missing data.
reasons for why they would not begin CBT-I, it was our impression that for most of those who declined, engaging in CBT-I seemed like too much effort.We also found that patients who declined participation in CBT-I (compared with those who completed it) were almost four years older.Among individuals eligible for clinical research studies of insomnia, refusal rates are quite variable.In some studies of CBT-I, refusal rates ranged from 26-45% (Sidani et al., 2010).
A limitation is that information was not systematically gathered from geriatricians regarding changes in their knowledge about, attitudes toward, and appraisal of the utility of CBT-I for their patients.This information was not obtained because this was a retrospective review of patient records.Formal implementation studies track indicators of systemic and clinician change (Bauer et al., 2015) and future implementation studies of CBT-I with older adults would benefit by formal measurement of these parameters.
A clinical challenge was that implementation of CBT-I with our older patients required adaptation for medical and functional issues that are less common in younger adults, even the "young old."Pain, frailty, and other common late-life morbidities may make engagement in CBT-I difficult.For example, one frail 90 year old patient could not do the stimulus control instruction to leave the bed if awake because he depended on his home health aide to leave his bed.We adapted the stimulus control instruction by mentally partitioning his bed: the left side was for sleep and right side for being awake.On the right was a bedside table on which was a book he would read until he was sleepy enough to move to the left "sleep side."This process has been referred to as counter-control (Zwart & Lisman, 1979).
As noted, another challenge was reluctance of almost half of eligible patients to participate in CBT-I.One way to address this would be greater effort on the part of geriatricians to build knowledge and enthusiasm among patients about CBT-I prior to referral for it.Written materials describing CBT-I and its efficacy might also be given to patients prior to referral.The option of receiving CBT-I by telehealth might increase participation rates especially among the "oldest old" for whom attending in-person sessions might be challenging.(Buysse et al., 2011).It was notable that the only characteristic that significantly differentiated treatment completers from nonparticipants was older age.

Treatment effectiveness
Overall, treatment effectiveness was excellent.Based on the ISI, two-thirds of patients were treatment responders and three-fifths were treatment remitters.These outcomes compare favorably to a large study of chiefly middle-aged adults in which ISI-defined response to an acute course of CBT-I was 59.9% and with a remission rate of 39.2% (Morin et al., 2009).Depending on the target sleep symptom, 68-81% of completers improved with CBT-I.Eighty-five percent of those who began treatment finished it.CBT-I treatment was generally completed within the eight session time frame of the CBT-I treatment manual (Perlis et al., 2008).Clinical studies, however, lack the rigor of randomized control clinical trials.A limitation of this report is that these older patients were predominantly white and well-educated and so it is unclear if CBT-I could be successfully and effectively implemented with other racial or socioeconomic older patient groups.However, CBT-I has been successful with varied patient populations of younger adults (Muench et al., 2022).Further, clinically meaningful outcomes have been reported in some randomized controlled trials of CBT-I that have included demographically diverse groups of older adults in the US Department of Veterans Affairs healthcare facilities (Alessi et al., 2016).One other potential limitation was that sleep parameters were self-recorded and not, for example, supplemented with actigraphy (i.e., objective measure of sleep).

Medication discontinuation
Over three-quarters of patients who were using sleep medications stopped taking them prior to beginning CBT-I; and one year later, 69% of them did not have a prescription in the EMR.Given the challenges of deprescribing sleep medications and the potentially adverse impact of them on older adults, we felt that this was a meaningful outcome (Kuntz, Kouch, Christian, Peterson, & Gruss, 2018).We could not discern, however, whether patients with an EMR documented sleep prescription were actually using them nor whether patients might have obtained sleep medications from another provider.Further, we did not examine whether patients who were not taking sleep medications at the beginning of CBT-I might have commenced taking them in the year following treatment.

Training : Foundation for effective implementation of CBT-I
Formal training in CBT-I is the foundation of successful implementation of CBT-I and other evidence-based treatments.Concerns exist about the limitations of traditional professional continuing education (Grus & Rozensky, 2019).Post-licensure acquisition of knowledge and skills to competently provide evidence-based treatment requires more than a brief workshop and skill development usually benefits from case consultation (Demiris, Parker, Oliver, Capurro, & Wittenberg-Lyles, 2014).As noted, there is a very limited number of individuals who have been formally trained in CBT-I, but training venues exist (Thomas et al., 2016).Of note, is a program to build CBT-I competency among clinical staff in the US Dept of Veterans Affairs (Karlin, Trockel, Taylor, Gimeno, & Manber, 2013).

Conclusions and implications
• CBT-I can be successfully implemented in a geriatric primary care clinic with a group of predominantly "old old" and "oldest old" patients.
• Sleep medications can be deprescribed among those taking them; and most will not be taking them one year after completion of CBT-I.

Table 1 .
Deprescribing and individual CBT-I treatment effectiveness outcomes.

•
Meaningful and sustained improvement can be made for most older patients in insomnia sleep targets as well as in treatment response and treatment remission as judged by the ISI.•With formal training in CBT-I, geropsychologists can play a key role in the treatment of late life insomnia.