Association between subjective–objective discrepancy of sleeping time and health-related quality of life: a community-based polysomnographic study

Abstract

Objective:

This study aimed to investigate the existence of a difference in quality of life (QOL) between individuals with and without significant subjective-objective discrepancy (SOD) in total sleep time (TST).

Methods:

From the Sleep Heart Health Study (SHHS) 2, 2540 individuals who had completed polysomnography, a morning sleep survey, and the 36-item Short-Form Health Survey (SF-36) were included in the analyses. The participants were classified as normoestimators (estimation of TST < ± 60 min), underestimators (underestimation of TST ≥ 60 min), or overestimators (overestimation of TST ≥ 60 min). The standardized SF-36 QOL scores were compared among the three groups. An adjusted partial correlation analysis was conducted between SOD and QOL.

Results:

Of the 2540 participants, 1617 (63.7%), 433 (17.0%), and 490 (19.3%) were assigned to the normoestimator, underestimator, and overestimator groups, respectively. The bodily pain and social functioning components of the SF-36 score were significantly lower in the underestimators than in the normoestimators, whereas the physical functioning component was significantly lower in the overestimators than in the normoestimators. The absolute value of SOD in the TST showed a significant negative correlation with the physical and mental components of the SF-36.

Conclusions:

QOL was significantly better in the normoestimator than in the other groups and linearly correlated with the absolute value of SOD. This study suggests that a high prevalence of positive and negative sleep misperception in a community population can be a potential factor associated with poor QOL and potential comorbidities.

Brief Summary

This is the largest study to date, comprising 2540 community-based participants with polysomnographic data, examining sleep under- and overestimation in relation to quality of life. More than one-third of participants estimated their sleeping time significantly more or less than the objective polysomnographic sleeping time. These sleep under- and overestimators, compared to normoestimators, reported a poor quality of life (physical and mental aspects), suggesting a substantial prevalence and mental health comorbidity of subjective-objective discrepancy in sleep.

Introduction

People subjectively perceive and estimate the quality, quantity, maintenance, and fragmentation of sleep. Subjective sleep reports can be quite different from objective sleep measurements using polysomnography (PSG) (1). Some individuals who sleep for >6–7 h may report that they never slept (2); others may report that they slept significantly more than their objective sleep time (3).

Patients with insomnia often underestimate their sleeping time (4-7). Sleep state misperception (SSM) refers to prevalent symptoms of insomnia without backup of objective measurement (8-10) and the International Classification of Sleep Disorders-2 defined SSM as paradoxical insomnia, one of the subtypes of primary insomnia (11). Underestimation of sleep frequently occurs in insomnia, but it is also common in the general population and in middle-aged adults (9, 12). Researchers have found that sleep underestimation also occurs in post-traumatic stress disorder (13), depressive disorder (14), chronic fatigue syndrome (15), and medical conditions such as rheumatoid arthritis, irritable bowel syndrome, and epilepsy (16-18). One study reported that 33% of obstructive sleep apnea (OSA) patients (n = 405) underestimated TST by >1 h compared to PSG (19). The etiology of sleep underestimation is unclear, but studies have shown that alterations in electroencephalography (20, 21) and neurophysiologic hyperarousal and fragmentation of the microstructure of sleep could influence sleep underestimation (22). A relatively less studied phenomenon is positive sleep estimation, in which insomnia patients overestimate their sleep duration (12, 23). Sleep overestimation seems to be prevalent, although there are no diagnostic criteria. Demographic studies have consistently shown that the general population tends to overestimate their sleep (12, 24, 25), especially in people with short sleep durations (26).

Individuals with sleep underestimation may experience poor health-related quality of life (QOL). It can also be assumed that there might be bi-directional associations between sleep underestimation and health-related QOL or that these may aggregate due to common risk factors. Clinical practice involving patients with sleep underestimation is quite challenging because conventional PSG may not reflect subjective sleep disturbances, resulting in under-treatment (27-29). Investigations on health-related QOL in association with both sleep under- and overestimation might provide information on the self-perceptive well-being of health outcomes in people with sleep misperception. Although studies have reported differences in psychiatric symptomatology or personality between sleep underestimation and control groups (30, 31), few have addressed the relationship between QOL and subjective-objective discrepancy in sleep (SODS). The Sleep Heart Health Study (SHHS) is an optimal study to investigate the relationship between SODS and QOL because it is a large, community-based, cohort study that obtained both subjective and objective sleep data as well as data on health-related QOL and other clinical factors.

The current study aimed to show the prevalence of subjective-objective discrepancy (SOD) of TST (defined as subjective TST in morning sleep survey minus objective TST in PSG) in a community-dwelling population aged >40 years; investigate differences in health-related QOL among normoestimator (−60 min < SODS < 60 min), underestimator (SODS ≤ −60 min), and overestimator (SODS ≥ 60 min) groups; and investigate the correlation between the difference in subjective and objective sleep duration and QOL score.

Materials and methods

Data sources and study cohort

This dataset is a large-scale open, multi-center, community-based, prospective cohort study that sought to determine the cardiovascular and other consequences of sleep-disordered breathing in the United States (ClinicalTrials.gov Identifier: NCT00005275) (32, 33). The current study primarily aimed to test whether sleep-related breathing is associated with an increased risk of hypertension, coronary heart disease, stroke, and mortality. From 1995 to 1998, 6441 men and women aged 40 years or older were enrolled to take part in the SHHS 1. From 2001 to 2003, a second study that included PSG data from 4586 SHHS participants (SHHS 2) was conducted. All participants provided written informed consent, and the study was approved by the institutional review board (IRB) of each participating institution. We used a subset of the SHHS 2 database obtained from the National Sleep Research Resource (https://sleepdata.org/datasets/shhs). We had access to the SHHS database by acquiring a signed agreement with Brigham and Women’s Hospital, and our project was exempted from review by the IRB of Gil Medical Center. This dataset includes data from a PSG study and morning survey about the previous night’s sleep after PSG were conducted, as well as health-related QOL that was assessed using the 36-item Short-Form Health Survey (SF-36). The analytical sample was restricted to those with complete PSG data, subjective TST in morning sleep survey data, and health-related QOL data from SF-36 scores.

Polysomnography

All participants underwent unattended, in-home, overnight PSG, as previously described using a Compumedics P-series recording system (Compumedics). The recording montage included a C3-A2 and C4-A1 electroencephalogram, left and right electrooculograms, chin electromyogram, single-lead electrocardiogram, airflow by an oral-nasal thermistor, oxyhemoglobin saturation by pulse oximetry, measurement of thoracic and abdominal effort by impedance plethysmography, and body position (by mercury strain gauge). Sleep stage scoring of all nocturnal recordings was conducted by trained technicians using the Rechtschaffen and Kales criteria at a centralized reading center (34). Apnea was identified if airflow was absent or nearly absent at ≥10 s (35). Hypopnea was identified and defined as a discernible reduction in airflow or thoracoabdominal movement (≥30% below baseline values) occurring for ≥10 s (35). These data were defined and scored using the apnea–hypopnea index (AHI) in various ways, and we chose the recommended hypopnea criteria of the American Academy of Sleep Medicine manual (10) (i.e., AHI was defined as the number of apneas or hypopneas, each associated with a 3% decrease in oxygen saturation per hour of sleep).

Calculation of SODS and group assignment

To calculate SOD of sleep duration, objective TST in PSG was subtracted from subjective TST as reported using the question, ‘How much time do you think you actually slept last night?’ in a morning sleep survey completed after the PSG. A difference of ≥60 min in the TST was defined as marked SODS (8, 26). Based on these data, participants were assigned into three groups: (1) normoestimator group (−60 min < SODS < 60 min), (2) underestimator group (SODS ≤ −60 min), or (3) overestimator group (SODS ≥ 60 min).

Quality of life data

QOL was determined using the SF-36, a widely used scale to determine health-related QOL. Each item is grouped into eight scales considered to be important in most health care situations: four scales (physical functioning, role-physical [role limitation due to physical problems], bodily pain, and general health) comprise the mental component score and the other four scales (vitality, social functioning, role-emotional [role limitations due to emotional problems and mental health]) comprise the physical component score. Domain and component standardized scores ranged from 0 to 100 according to the methods for computing scores, with higher scores indicating better QOL (36).

Medical illness, medication use, and insomnia

To identify medical diseases and medication use in the participants, we determined whether participants used antihypertensives, diabetes mellitus (DM) medications, antidepressants (tricyclic antidepressants and non-tricyclic antidepressants), and benzodiazepines. Medication use was evaluated using a standardized medication survey, as previously reported (37). DM was defined as current use of insulin or oral hypoglycemic agents, and hypertension was defined as the average of the second and third systolic blood pressure readings >140 mmHg or average diastolic blood pressure >90 mmHg, or current use of antihypertensive medications. Insomnia was evaluated using four items (“Have trouble falling asleep,” “Wake up during the night and have difficulty getting back to sleep,” “Wake up too early in the morning and be unable to get back to sleep,” and “Take sleeping pills or other medication to help you sleep”) with a 5-point Likert scale response (1, never; 2, rarely [1x/month or less]; 3, sometimes [2–4 x/month]; 4, often [5–15 x/month]; 5, almost always [16–30 x/month]). As formal diagnostic evaluations including assessing significant distress, daytime consequences, and symptom chronicity were not performed, participants were defined as having insomnia if any of the four items were answered 5 (almost always).

Statistical analysis

Demographic, clinical, and PSG data were compared across the three groups using analysis of variance. To compare QOL scores among the groups, we employed non-parametric analysis of covariance controlling for age, sex, TST in PSG, AHI, DM, and hypertension, and post-hoc testing for Bonferroni-adjusted multiple comparisons. To investigate correlations between the absolute value of SODS and QOL, we performed a non-parametric partial correlation analysis controlling for the same potential confounding factors. Additionally, we performed group comparisons and correlation analyses after adding the presence of insomnia and the use of antidepressants to the covariates. Statistical analysis was performed at a two-sided significance level of p < 0.05, using SPSS for Windows (version 23, IBM Corp.) and SAS Enterprise Guide version 9.4 (SAS Institute, Inc.).

Results

In total, 2540 participants were included in the final study population, of which 1617 individuals were normoestimators (63.7%), 433 were underestimators (17.0%), and 490 were overestimators (19.3%) (Figure 1).

Figure 1.

Flow diagram of the study population

Participant characteristics and polysomnographic data

Table 1 shows the demographic and clinical characteristics and the PSG results. The average age was 62.4 years; 1169 (46.0%) participants were male. The mean subjective TST on the morning sleep survey was 378.1 min and mean SOD of TST of all participants was 4.3 min. In PSG, the mean sleep efficiency was 79.2%, the mean AHI was 16.5 per hour of sleep, and 39.7% of participants had an AHI value of ≥15. Of the total participants, 36.3% had a significant SODS.

Table 1.

Demographic, clinical, and polysomnographic data of participants

Variable	n = 2540
Demographics
Age (years)	62.4 ± 10.5
Sex (male)	1169 (46.0%)
Body mass index (kg/m2)	28.3 ± 5.1
Sleep habit and subjective total sleep time
Epworth sleepiness scale	7.5 ± 4.2
Subjective total sleep time in morning survey after polysomnography (minutes)	378.1 ± 86.6
Polysomnographic data
Sleep and wake time
Time in bed (minutes)	474.3 ± 70.6
Total sleep time (minutes)	373.8 ± 69.7
Sleep latency (minutes)	25.8 ± 30.1
Sleep efficiency (%)	79.2 ± 11.7
WASO (minutes)	80.2 ± 55.1
REM sleep latency (minutes)	96.9 ± 62.3
Respiration
AHI (events per hour)*	16.5 ± 16.1
AHI ≥ 15	1008 (39.7%)
Arousal index	18.4 ± 10.2
Subjective–objective discrepancy of sleep
SOD of TST (minutes, TST in morning survey minus TST in PSG)	4.3 ± 79.0
Normoestimator (−60 m < SOD < 60 m)	1617 (63.7%)
Underestimator (SOD ≤ −60 m)	433 (17.0%)
Overestimator (SOD ≥ 60 m)	490 (19.3%)

Data are presented as mean ± standard deviation or number (%).

WASO, wake time after sleep onset; REM, rapid eye movement; AHI, apnea–hypopnea index; SOD, subjective–objective discrepancy; TST, total sleep time; PSG, polysomnography

^*AHI: number of apneas and hypopneas with ≥3% oxygen desaturation per hour of sleep

Comparison of clinical and PSG data among the groups

Table 2 shows clinical features and PSG results of three groups according to sleep estimation and their inter-group differences. In the overestimator group, age and the proportion of men were higher (p < .001) than in either of the other groups. A significant difference in the frequency of hypertension (p < .001) and DM (p = .002) was observed among the three groups, with higher proportions in the overestimator group as compared to either of the other groups. There were no significant differences in body mass index or frequency of tricyclic antidepressant, non-tricyclic antidepressant, and benzodiazepine use during the past 2 weeks across groups (ps > .05). The underestimator group showed the highest proportion of insomnia, highest proportion of frequent (>15 days/month) insomnia symptoms (p < .001).

Table 2.

Comparison of demographic and PSG data among the groups

Variable	Under estimator* (n=433)	Normo estimator (n=1617)	Over estimator (n=490)	Statistics ‡
Demographics
Age, years	61.4 ± 10.8	61.3 ± 10.4	67.2 ± 9.6	F = 63.83, p < .001
Sex, male	39.0%	43.9%	59.2%	χ2 = 45.60, p < .001
Body mass index, kg/m2	28.1 ± 5.3	28.4 ± 5.0	27.9 ± 5.0	F = 2.04, p = .13
Subjective and objective TST
Subjective TST, min	286.0 ± 78.7	382.5 ± 68.7	446.7 ± 75.1	F = 601.28, p < .001
Objective TST, min	400.3 ± 61.6	378.5 ± 64.9	336.3 ± 75.5	F = 119.71, p < .001
SOD of TST, min	−114.4 ± 53.5	4.0 ± 31.5	110.4 ± 53.2	F = 3582.36, p < .001
Insomnia
Insomnia > 15 days/month (yes), %	20.1%	11.6%	9.5%	χ2 = 28.50, p < .001
Medical illness and medications
Hypertension, %	51.7%	51.1%	62.0%	χ2 = 18.66, p < .001
Diabetes mellitus medication, %	7.0%	8.4%	13.1%	χ2 = 12.87, p = .002
Antidepressant TCA, %	1.4%	2.5%	3.7%	χ2 = 4.91, p = .086
Antidepressant Non-TCA, %	9.75%	8.0%	8.8%	χ2 = 1.43, p = .50
Benzodiazepines, %	4.9%	5.4%	5.9%	χ2 = 0.50, p = .78
Polysomnographic data
Sleep architecture, %
Stage N1	5.7 ± 3.7	5.3 ± 3.5	7.3 ± 5.4	F = 45.78, p < .001
Stage N2	57. 2 ± 10.4	57.2 ± 10.9	59.7 ± 11.9	F = 9.24, p < .001
Stage N3	16.5 ± 10.7	16.5 ± 10.8	13.9 ± 11.4	F = 10.80, p < .001
Stage R	20.6 ± 6.2	21.0 ± 6.3	19.1 ± 7.1	F = 14.66, p < .001
Sleep and wake time
Time in bed, min	498.2 ± 64.9	469.4 ± 66.8	469.5 ± 82.3	F = 30.58, p < .001
TST, min	400.1 ± 61.6	378.3 ± 65.2	335.5 ± 75.2	F = 118.30, p < .001
Sleep latency, min	19.2 ± 26.8	19.5 ± 26.9	24.0 ± 33.7	F = 5.33, p = .005
Sleep efficiency, %	80.6 ± 9.5	80.9 ± 10.8	72.1 ± 13.6	F = 122.97, p < .001
WASO, min	78.9 ± 49.5	71.5 ± 48.8	110.0 ± 67.6	F = 98.76, p < .001
REM sleep latency, min	101.3 ± 67.5	94.3 ± 57.6	102.0 ± 72.0	F = 3.71, p = .25
Respiration
AHI, event per hour	15.3 ± 15.0	15.9 ± 15.5	19.5 ± 18.5	F = 11.02, p < .001
AHI ≥ 15, %	35.6%	38.9%	45.9%	χ2 = 11.44, p = .003
Arousal index	18.0 ± 9.0	17.7 ± 9.5	20.9 ± 12.6	F = 18.56, p < .001

Data are mean ± standard deviation (SD) or percent (%) values.

PSG, polysomnography; TST, total sleep time; SOD, subjective–objective discrepancy; TCA, tricyclic antidepressant; Non-TCA, non-tricyclic antidepressant; WASO, wake time after sleep onset; REM, rapid eye movement; AHI, apnea–hypopnea index.

^*Participants were classified as underestimators (underestimation of TST ≥ 60 min), normoestimators (estimation of TST < ± 60 min), and overestimators (overestimation of TST ≥ 60 min).

^‡Analysis of variance or χ² test

The overestimator group had the highest proportion of stage N1 (p < .001) and N2 (p < .001) and the lowest proportion of stage N3 (p < .001) and R (p < .001) among the groups in PSG sleep architecture. They had the lowest sleep efficiency (p < .001), longest sleep latency (p < .001), highest wake after sleep onset (WASO; p < .001), and highest AHI (p < .001) among all the groups. In contrast, the underestimator group had the longest PSG-assessed TST (p < .001) and time in bed (p < .001) among all the groups.

Comparisons of SF-36 scores among the groups

The SF-36 standardized scores were compared among the three groups with adjustment for potential confounding factors (i.e., age, sex, TST, AHI, hypertension, and DM), which were significantly different among the groups (Table 3). The adjusted values for the physical functioning score were significantly lower in the overestimator group than in the normoestimator group (p = .024). Social functioning scores were significantly lower in the underestimator group than in the normoestimator group (p = .037). The bodily pain score was lower in the underestimator group than in the normoestimator group (p = .014) and the overestimator group (p = .046). These differences remained significant after bootstrapping (Table S1, Supplemental Digital Content). Comparisons of SF-36 scores with additional adjustment for the presence of insomnia and the use of antidepressants showed no significant differences among the groups in physical functioning (p = .050) or bodily pain (p = .052). The results are listed in Table S2. Comparisons of SF-36 scores among groups according to WASO and SWS are also presented in the Supplemental Digital Content (Tables S3 and S4).

Table 3.

Comparison of standardized scores in SF-36 among the groups

SF-36 standardized scores	Under estimator*	Normo estimator	Over estimator	Statistics ‡		Post-hoc analysis§
				F	p	Normo vs. Over	Normo vs. Under	Over vs. Under
Physical components
Physical functioning	77.8 ± 23.9	78.8 ± 23.6	71.7 ± 26.3	3.26	.039	.024	.91	.33
Role physical	74.9 ± 36.7	78.6 ± 34.2	68.7 ± 39.6	3.68	.025	.064	.10	>.99
Bodily pain	69.3 ± 23.5	72.6 ± 22.4	72.0 ± 23.2	4.03	.018	.99	.014	.046
General health	72.2 ± 18.6	73.4 ± 18.3	70.3 ± 19.8	1.83	.16	.91	.13	.60
Physical component summary score	46.1 ± 10.6	47.1 ± 10.1	44.6 ± 11.0	2.11	.12	.51	.13	.90
Mental components
Vitality	63.4 ± 19.4	63.7 ± 18.4	62.2 ± 20.0	0.11	.90	>.99	.95	.99
Social functioning	89.3 ± 18.8	91.8 ± 15.9	89.5 ± 18.5	3.06	.047	.53	.037	.65
Role emotional	93.0 ± 18.0	93.3 ± 17.8	91.5 ± 18.8	2.19	.11	.085	.96	.48
Mental health	79.7 ± 15.3	81.6 ± 13.4	81.4 ± 15.1	1.47	.23	>.99	.21	.46
Mental component summary score	54.2 ± 7.8	54.8 ± 7.5	55.0 ± 8.4	0.84	.43	.93	.62	.50

Data are mean ± standard deviation (SD) or n (%) values.

SF-36, short-form 36 health survey; Normo, normoestimator; Over, overestimator; Under, underestimator

^*Participants were classified as underestimators (underestimation of total sleep time ≥ 60 min), normoestimators (estimation of total sleep time < ± 60 min), and overestimators (overestimation of total sleep time ≥ 60 min).

^‡Analysis of covariance controlling for age, sex, total sleep time, apnea–hypopnea index, hypertension, and diabetes mellitus.

^§P values of post-hoc tests for Bonferroni-adjusted multiple comparisons.

Significant results are highlighted in bold.

Correlation among the difference of SODS and QOL

Because QOL scores differed in both the underestimator and overestimator groups relative to the normoestimator group, we then assessed the partial correlation between the absolute values of SODS and QOL scores, adjusted for age, sex, TST, AHI, hypertension, and DM (Table 4). Physical functioning (p = .042), role-physical (p = .015), general health (p = .039), physical component summary score (p = .039), and social functioning (p = .046) each negatively correlated with the absolute value of sleep estimation (rho: from −0.01 to −0.05). When the presence of insomnia and the use of antidepressants were added to the covariates, only role-physical (p = .046) correlated with the absolute value of sleep estimation (Table S5, Supplemental Digital Content).

Table 4.

Partial correlation among the absolute value of SOD^* of total sleep time and SF-36 scores

SF-36 standardized scores	rho ‡	p value
Physical components
Physical functioning	−0.04	.042
Role physical	−0.05	.015
Bodily pain	−0.03	.092
General health	−0.04	.039
Physical component summary score	−0.04	.039
Mental components
Vitality	−0.01	.78
Social functioning	−0.04	.046
Role emotional	−0.03	.18
Mental health	−0.03	.16
Mental component summary score	−0.01	.71

SOD, subjective–objective discrepancy; SF-36, short-form 36 health survey

^*SOD, subjective total sleep time in morning survey minus objective total sleep time in PSG

^‡Partial correlation analysis was conducted controlling for age, sex, total sleep time, apnea–hypopnea index, hypertension, and diabetes mellitus.

Significant results are highlighted in bold.

Discussion

The present study, with large community-based polysomnographic data, is one of the first studies to investigate the concept of both sleep under- and overestimation in relation to QOL. With more than one-third of the study sample reporting SODS, the results of this study support the clinical significance of SODS in community populations and its relationship with subjective well-being. We found that QOL was lower in individuals who either underestimated or overestimated their sleep duration relative to PSG, compared to individuals with closer concordance between the perceived and measured sleep durations. Moreover, there was a negative linear association between the degree of SODS in either direction (under- or overestimation) and QOL.

The prevalence of SODS in this large, community-based sample of older adults was 36.3% (overestimators 19.3%; underestimators 17.0%). Because of the lack of previous studies, there is no consensus on the prevalence of SOD in TST in the general population, although the prevalence of sleep underestimation in previous studies ranged from 9.2% to 50% in insomnia patients (5, 38-40). The high prevalence of SODS in this large sample suggests that it is more common than diagnosed insomnia, although likely underappreciated (41).

We found that the underestimator group was younger, more likely to be female, and had a lower prevalence of medical illnesses than the other groups. Importantly, the underestimators had a lower QOL than the normoestimators in both the physical and mental components, including bodily pain and social functioning. These results may imply that underestimators might have sleep misperceptions due to poor QOL, or vice versa (42). However, the higher proportion of insomnia in the underestimator group in our results might also account for the low health-related QOL in the underestimators (43). We also found a significant association between sleep underestimation and QOL even after adjusting for the presence of insomnia and the use of antidepressants. There has been a suggestion of the possibility of common aggregates of low QOL and sleep underestimation due to other common causes such as psychiatric problems, which can exaggerate sleep deficits and complaints (9). Insomnia patients with underestimation were found to have a higher level of neuroticism (30), hysteria (31), hypochondriasis, hysteria, and psychasthenia (44) than objective insomnia or healthy controls. Conversely, insomnia patients with sleep underestimation tended to show lower anxiety scores (45) and comparable levels of depression, anxiety, and QOL in the SF-36 compared with insomnia patients without SSM (40). Although cautious interpretation is needed, underestimation can be related to low health-related QOL in both physical and mental aspects in association with insomnia and depression.

The overestimator group was older, showed lower symptoms of insomnia, higher AHI, and a lower proportion of stage N3 and REM sleep than the other groups. These findings are consistent with previous reports that sleep overestimation might be associated with sleep apnea, snoring, daytime sleepiness, and altered sleep stages (3, 23, 46, 47). Some OSA patients overestimate TST because they paradoxically fall asleep instantly because of excessive sleepiness; therefore, they are often unaware of their sleep disruption compared with insomnia patients (48-50). Additionally, the SHHS cohort is different from previous clinic-based cohorts in that the participants were older and recruited from the community, thus including many undiagnosed individuals with OSA. A higher proportion of moderate to severe OSA in this population (40% ≥ AHI=15) than in previous studies (0%–50% ≥ AHI=15) (51-53) might have resulted in a high proportion of overestimators. We also found that QOL scores were lower in the overestimator group than in the normoestimator group. Due to their lower physical function, individuals with overestimation might have experienced limitations in performing work or daily physical activities (42) despite their sufficient amount of subjective sleep. Notably, the overestimator group showed a lower QOL only in the physical components. Positive SSM is a relatively new concept to the previously described sleep underestimation (23), and little is known about the relationship between sleep overestimation and QOL or psychiatric symptoms. In this study, fewer insomnia symptoms were found in the overestimator group than in the other groups. Patients with sleep disorders with sleep overestimation previously showed moderate daytime sleepiness in multiple sleep latency tests because of a shortage of objective sleep, but most are not aware of their lack of sleep or sleep disruption and tend to overestimate TST (23). This paradoxical relationship between sleep overestimation and sleep complaints may explain our finding of poor QOL in the physical components of the overestimators (23). Additionally, a linear correlation was found between the absolute value of SODS and QOL in components such as physical functioning, role-physical, general health, physical component summary score, and social functioning. Altogether, our results suggest that sleep underestimation and overestimation are potentially associated with a low QOL in mental and physical components, respectively.

A notable strength of our study is that we used one of the largest community cohort PSG datasets relating SODS to QOL. The only previous study to investigate whether QOL differs between patients with insomnia with and without SSM did not show a significant difference in SF-36 scores, which is inconsistent with the findings of the current study (40). Acquisition of both PSG and subjective sleep estimation data on the next morning sleep survey permitted a more reliable definition of sleep under-, normo-, and overestimation because some of those previous studies defined subjective insomnia based only on PSG data (26, 40). Additionally, previous studies investigating both subjective and objective sleep data had relatively small sample sizes (21–866 participants) and different characteristics of the study population (17, 26, 30, 40, 44, 45). These sample size issues and heterogeneity of previous studies might have resulted in inconsistent results (30, 31, 40, 44). Collectively, our study, in the largest community-based sample, the use of both subjective and objective measures of TST, and a variety of additional variables, might have indicated the association between under-, normo-, and overestimators and health-related QOL.

However, there were several limitations to the present study. This cohort was relatively old (mean age: 62.4±10.5) with a high prevalence of moderate or severe OSA. Since one of the aims of SHHS was to find the association between sleep-disordered breathing and cardiovascular diseases (32), the characteristics of the study population could bias the results from generalizing to other populations. Additionally, the effect of insomnia disorder on sleep misperception, QOL, or their association may differ from what would be observed in a clinical setting given discrepancies in the criterion used to classify insomnia in this study (insomnia symptoms prevailed for more than 15 days per month). Another limitation of our study is the lack of longitudinal evaluation, which prevented us from inferring the causal directions between SODS and health-related QOL.

Conclusion

These novel findings in a community-based cohort with PSG data showed a higher prevalence of SODS and better QOL in participants without SODS than in those with SODS. These findings support the idea that SODS, assessed as under- or overestimation of sleep duration, is associated with poor QOL and high comorbidities. Our data suggest that SODS is a phenotype that should be further investigated for its epidemiologic characteristics and associated psychiatric and cognitive consequences, in studies of both patients with sleep disorders and the general population. Previous imaging studies have found brain alterations in patients with insomnia and SODS using methods such as sleep electroencephalography (20, 21), event-related potentials (8, 54), and cerebral glucose metabolism (55). Future neuroimaging studies might reveal the neurobiology of SODS in people with neuropsychiatric disorders and the general population. Since the correlation between SODS and QOL varies, future studies should also focus on more refined designs with thorough confounding factors and longitudinal designs.

Abbreviations:

QOL

quality of life

SOD

subjective-objective discrepancy

SODS

subjective-objective discrepancy in sleep

TST

total sleep time

SHHS

Sleep Heart Health Study

PSG

polysomnography

SSM

sleep state misperception

OSA

obstructive sleep apnea

AHI

apnea–hypopnea index

DM

diabetes mellitus