Volume 69, Issue 10 p. 2993-2995
Research Letter
Free Access

Efficacy of cognitive behavioral therapy for insomnia in geriatric primary care patients

Gregory A. Hinrichsen PhD

Corresponding Author

Gregory A. Hinrichsen PhD

Department of Geriatrics and Palliative Medicine, Icahn School of Medicine at Mount Sinai, New York, New York, USA


Gregory A. Hinrichsen, PhD, Department of Geriatrics and Palliative Medicine, Icahn School of Medicine at Mount Sinai, One Gustave L. Levy Place, Box 1070, New York, N.Y. 10029.

Email: [email protected]

Search for more papers by this author
Rosanne M. Leipzig MD, PhD

Rosanne M. Leipzig MD, PhD

Department of Geriatrics and Palliative Medicine, Icahn School of Medicine at Mount Sinai, New York, New York, USA

Search for more papers by this author
First published: 13 June 2021
Citations: 4

Funding information: Fan Fox and Leslie R. Samuels Foundation (to Rosanne Leipzig)


Insomnia is common in older adults and associated with increased mortality and morbidity including depression, anxiety, cognitive decline, and increased risk for falls.1 Prevalence rates range from 12% to 41%,2 and are greatest in those with medical problems. Most medications for insomnia are listed in the American Geriatrics Society Beers Criteria as potentially inappropriate for use in older adults due to increased adverse events including falls, fractures, and motor vehicle accidents.3 Cognitive behavioral therapy for insomnia (CBT-I) is an effective non-pharmacological treatment for insomnia, and is recommended as the first-line treatment by the American College of Physicians and other professional organizations,4 noting concern about long-term use of hypnotics. CBT-I has been found to be effective in the treatment of insomnia in older adults but most study participants are between 65 and 74 years.5 It is not known whether CBT-I would be acceptable and effective in the “old” (aged 75–84) and the “oldest old” (85 years and older) who often have more medical and psychosocial concerns than those in their 60s and early 70s.


Patient population

Subjects were patients of the Icahn School of Medicine at Mount Sinai geriatric primary care practice who, during the years 2015–2018, were referred by their geriatrician to GAH, a clinical geropsychologist, for assessment of insomnia and treatment with CBT-I. The practice has 4500 patients whose median age was 85 years. Thirty-four referred patients were judged appropriate for CBT-I because they met the Diagnostic and Statistical Manual of Mental Disorders (DSM-5) criteria for insomnia, did not have other sleep disorders (including apnea), were cognitively intact, and were interested in taking part in the treatment. Of the 34 patients, 29 (85%) who began CBT-I completed it.


CBT-I was guided by the treatment manual, Cognitive Behavioral Treatment of Insomnia: A Session-by-Session Guide.6 Participants had an initial evaluation and then 1–2 weeks of baseline sleep diary data were completed and reviewed. After this, the mean number of CBT-I sessions was 6.1 (SD = 3.3, median = 5). Typically, treatment was ended when the patient evidenced 2 weeks of sustained improvement based on sleep diaries. Some patients had lengthier treatment because they did not evidence sustained improvement, were not satisfied with the degree of improvement, and/or a plateau was reached. This pre- and post-treatment study of clinical records was approved by the Icahn School of Medicine at Mount Sinai IRB.


Demographic and medical date were gleaned from the electronic medical record (EMR). Four validated instruments, the Patient Health Questionnaire (PHQ-9),7 a 9-item depression screen, the Generalized Anxiety Disorder 7-item scale (GAD-7),8 an anxiety screen, the Insomnia Severity Index (ISI),9 a commonly used patient report of insomnia severity, and the Epworth Sleepiness Scale (ESS),10 a patient report measure of daytime sleepiness, were administered to each patient at the initial interview and completion of CBT-I. Sleep diaries yielded continuous data on sleep onset latency (SOL; minutes to fall asleep), wake after sleep onset (WASO; minutes awake after falling asleep), early morning awakening (EMA; minutes awake earlier than intended), total sleep time (TST), and sleep efficiency (the percentage of intended sleep time actually slept).

Data analysis

Only those 29 patients who completed a full course of CBT-I were included in pre–post analyses where sleep parameters, PHQ-9 (without the sleep item, hence renamed PHQ-8), GAD-7, ISI, and ESS differences, were compared using paired t tests. The magnitude of the effect size of these outcomes was estimated using Cohen's d (Table 1).

TABLE 1. CBT-I completers changes in sleep, depression, and anxiety parameters (N = 29)
Outcome Baseline End of treatment Paired t test (df), p Effect size Cohen's d
Mean (SD) Mean (SD)
ISI 16.3 (5.2) 5.7 (4.9) 11.6 (26), <0.001 2.2
ESS 6.4 (4.7) 4.9 (4.4) 2.6 (28), 0.013 0.49
PHQ-8 4.2 (3.7) 2.3 (2.5) 2.5 (25), 0.020 0.49
GAD-7 5.9 (4.8) 2.9 (2.9) 3.7 (26), 0.001 0.71
SOL 46.0 (32.3) 21.9 (21.7) 4.8 (27), <0.001 0.91
WASO 41.4 (36.1) 14.4 (11.8) 4.2 (27), <0.001 0.80
EMA 29.3 (28.4) 10.5 (10.7) 3.8 (27), 0.001 0.73
TST 368.4 (58.9) 358.0 (54.2) 1.3 (27), 0.205 0.25
SE 75.0 (11.5) 88.0 (6.9) −7.7 (27), <0.001 1.4
  • Note: Possible ranges of measures: ISS (0–28); ESS (0–24); PHQ-8 (0–24); GAD-7 (0–21); SE (range 0–100%). Cohen's d effect size: d = 0.20, small; d = 0.5, medium; d = 0.8, large; d = 1.2 very large; d = 2.0, huge. For some measures, total N is not 29 because of missing data.
  • Abbreviations: CBT-I, cognitive behavioral therapy for insomnia; df, degrees of freedom; EMA, early morning awakening; ESS, Epworth Sleepiness Scale; GAD-7, Generalized Anxiety Disorder 7-item scale; ISI, Insomnia Severity Index; PHQ-8, Patient Health Questionnaire (8 of 9 items); SE, sleep efficiency; SOL, sleep onset latency; WASO, wake after sleep onset; TST, total sleep time.


Background characteristics of completers

CBT-I completers were predominantly white, well-educated women with insomnia of 5 years duration or longer, who had multiple medical problems and medications listed in the EMR. Mean age was 77.2 years, with almost two-thirds of patients aged between 76 and 93 years (Table S1).

Treatment effects

Table 1 summarizes pre–post changes in sleep parameters and mood/sleep questionnaires among completers. TST remained unchanged, however, there was a significant improvement in all other measures. The largest effect was a decrease in the severity of insomnia, followed by markedly improved sleep efficacy, and a large decrease in the duration of SOL and WASO. Sleepiness, depression, anxiety, and EMA also decreased significantly.


Frailty, pain, nocturia, cognitive limitations, and late evening/early morning medical/medication regimens can interfere with sleep especially among very old patients and may make insomnia more difficult to treat. Nonetheless, among a group of predominantly old-old and oldest old patients, we found that CBT-I is a highly effective treatment for chronic insomnia. Findings were similar to those of a randomized clinical trial of behavioral treatment of chronic insomnia that was briefer than described in this report (four sessions), delivered by a non-sleep specialist (in person and then telephone sessions) to a group of older adults who were almost 6 years younger than those in our study.11 The strengths of this study include participation of individuals older than in previous CBT studies, and a model of care providing colocation and partnership between primary care geriatricians and a geropsychologist. Limitations are that patients were predominantly white and well-educated and it is unclear whether CBT-I would be of interest or effective in other patient populations of advanced age.


We thank Michael Perlis (PhD) and Donn Posner (PhD) for training Gregory A. Hinrichsen in CBT-I, and Donn Posner (PhD) for his skillful consultation on initial CBT-I cases. This project was partially funded by the Fan Fox and Leslie R. Samuels Foundation.


    The authors have no conflicts of interest to report.


    Both authors significantly contributed to this manuscript. Gregory A. Hinrichsen and Rosanne M. Leipzig designed the protocol and supervised its progress, and drafted the manuscript. Gregory A. Hinrichsen conducted the data analysis.