A Single Item Screen for Clinically Significant Insomnia Symptoms in Community-Living Older Adults

INTRODUCTION

Insomnia symptoms (IS) are common among community-living older adults and associated with adverse outcomes,¹ but more efficient screening methods are needed.² We compared a single sleep-related item from the Center for Epidemiologic Studies Depression Scale (CESD-SI)³ to a validated measure of IS.^4,5 We hypothesized that the CESD-SI would show acceptable performance as a screen for clinically significant IS and be significantly associated with conditions known to be associated with poor sleep.¹

METHODS

Data came from the Precipitating Events Project.⁶ We included participants who completed sleep questionnaires during the 2005-2007 assessment. Of the 754 original participants, 287 died, 25 withdrew, and 63 were excluded due to living in a nursing facility, requiring a proxy, or not completing insomnia or depression questionnaires. Hence, the analytical sample included 356 participants.

We examined characteristics, defined in the footnotes of Table 1, including self-rated health status, number of chronic conditions, number of medications, cognitive impairment, functional impairment, restless leg symptoms, use of a sleep medication, and excessive daytime sleepiness.⁷ The CESD is an 11-item measure of depression.³ The sleep item (CESD-SI), “my sleep was restless”, asks participant how often they felt this way in the past week: “rarely or never” (1), “some of the time” (2), or “much of most of the time” (3). The Insomnia Severity Index (ISI; range 0-28) is a validated measure of IS, including seven items characterizing symptoms and consequences of insomnia.^4,5

Table 1.

Sample characteristics overall and by presence of clinically significant insomnia symptoms^*

Characteristic	Overall N=356	CESD-SI <2 N=131	CESD-SI ≥2 N=225	p value a
	Mean (SD) or n/N (%)
Age	84.3 ±4.6	84.3 ±0.39	84.4 ±0.31	0.90
Female sex	244 (68.5)	79 (60.3)	165 (73.3)	0.01
Minority race/ethnicity	41 (11.5)	19 (14.5)	22 (9.8)	0.18
Did not complete high school	108 (30.3)	34 (26.0)	74 (32.9)	0.17
Self-rated health status is fair or poor	119 (33.4)	31 (23.7)	88 (39.1)	0.003
Number of chronic conditions b	2.3 ±1.2	2.2 ±0.1	2.3 ±0.1	0.32
Number of medications c	9.2 ±3.8	8.9 ±0.4	9.4 ±0.2	0.26
Cognitive impairment d	54 (15.2)	23 (17.6)	31 (13.8)	0.34
Functional impairment e	3.2 ±3.5	3.3 ±3.1	3.6 ±3.3	0.32
Restless legs symptoms f	130 (40.8)	29 (22.1)	101 (44.9)	<0.001
Use of a sleep medication g	109 (30.6)	23 (17.6)	86 (38.2)	<0.001
Excessive daytime sleepiness h	76 (22.6)	27 (20.6)	49 (21.8)	0.04
ISI score	6.9 ±5.5	2.4 ±0.2	9.6 ±0.3	<0.001
ISI ≥8	153 (42.9)	6 (4.6)	147 (65.3)	<0.001
CES-D total score	11.5 ±9.3

Abbreviations: CESD= Center for Epidemiological Studies-Depression Survey; CESD-SI=CESD sleep item; ISI=Insomnia Severity Index.

^*Defined by CESD sleep item score ≥2

^aDerived from t-tests for continuous and chi-square test (df = 1) for categorical variables. A p value < 0.05 indicated statistical significance.

^bNumber of self-reported, physician-diagnosed medical conditions, including hypertension, myocardial infarction, heart failure, stroke, cancer (excluding minor skin cancer), diabetes, hip fracture, arthritis, and/or chronic lung disease.

^cPrescribed and over-the-counter medications

^dMini Mental Status Examination score <24

^eDisability in 0-12 basic, instrumental, and mobility activities

^fSelf-reported feeling of restlessness in the legs at night, while lying in bed, at least one to two times per week over the previous two weeks.

^gAnswered yes to “in the past two weeks, have you taken a sleep aide to help you sleep during the night?”

^hEpworth Sleepiness Scale score ≥10. See reference 7 for justification.

The Pearson correlation coefficient compared the CESD-SI and total ISI score. Using cut-offs for the CESD-SI of ≥2 and 3, we calculated the sensitivity, specificity, and positive and negative predictive values for an ISI score ≥8 (i.e., the threshold for clinically significant IS).^4,5 Using chi-square or t-tests, we evaluated the associations between conditions known to be associated with poor sleep¹ and clinically significant IS, defined by a CESD-SI score of ≥2, using STATA version 17 (StataCorp 2022; College Station, TX).

RESULTS

Characteristics of the cohort are shown in Table 1. The CESD-SI and ISI score were highly correlated (correlation coefficient=0.74; p <0.001). Participants with a score of 1 on the CESD-SI had a median ISI score of 2 (interquartile range [IQR] 0-4), while those with scores of 2 and 3 had median ISI scores of 7 [4-10] and 13 [10-16], respectively. A CESD-SI score ≥2 had the best sensitivity and negative predictive value (Table 2). Females and persons with poor self-rated health, restless leg symptoms, use of a sleep medication, and daytime sleepiness (ESS score ≥10) were significantly more likely to have clinically significant IS (p values <0.05).

Table 2.

Sensitivity, specificity, and predictive values of CESD-SI scores of ≥2 or 3 for detecting an ISI score of ≥8

Score on CESD-SI	Sensitivity	Specificity	Positive predictive value	Negative predictive value
≥2 (restless sleep some or most of the time)	96%	62%	65%	95%
3 (restless sleep most of the time)	51%	96%	91%	72%

Abbreviations: CESD-SI= Center for Epidemiological Studies Depression Survey sleep item; ISI=Insomnia Severity Index.

DISCUSSION

We compared the performance of a single item from the CESD to detect clinically significant IS using the ISI, a well-validated insomnia questionnaire, in community-living adults with advanced age. A CESD-SI score ≥2 had a high sensitivity and negative predictive value when compared to an ISI score ≥8, and persons with scores ≥2 were more likely to have conditions known to be associated with poor sleep.¹ We have expanded the results of previous studies^8,9 by demonstrating the performance of the CESD-SI in a community-living population that included both men and women with advanced age and through comparison with a reliable and valid insomnia instrument.^4,5

A CESD-SI of ≥2 performed best as a screen for clinically significant IS. A sensitivity of 96% is optimal to minimize false negatives. The high negative predictive value suggests that busy clinicians engaged in the care of older adults with multiple competing priorities can be reassured that no further evaluation of IS is needed for persons without restless sleep. The modest specificity for this threshold may be due to sleep conditions other than insomnia (e.g., sleep apnea, circadian rhythm disorders). Because these conditions are also important to detect and treat, prioritizing sensitivity is optimal.

Our analysis was limited to a mostly-white cohort from Connecticut. Also, we compared the CESD-SI to an insomnia instrument, not a diagnostic interview for insomnia disorder. A positive screen on the CESD-SI does not confirm a diagnosis of insomnia disorder, which requires demonstration of dissatisfaction with sleep quality and/or quantity (including difficulty initiating or maintaining sleep or waking too early) occurring ≥3 nights a week for ≥3 months with a negative impact on daytime functioning, as well as ensuring adequate sleep opportunity and ruling out other sleep disorders.¹⁰ Rather, the single item represents a feasible method to detect concerning sleep symptoms and should be followed by a detailed history to ascertain the underlying etiology of the IS.

In conclusion, the CESD-SI has excellent sensitivity to detect clinically significant IS in community-living adults with advanced age. Future research should compare the CESD-SI to gold-standard measures for detecting insomnia disorder. Use of this item in clinical and research settings could increase detection of concerning sleep symptoms and improve patient-centered care in this high-risk population.