Observation and Interview-based Diurnal Sleepiness Inventory for measurement of sleepiness in patients referred for narcolepsy or idiopathic hypersomnia

Abstract

Study Objectives:

First, to determine whether the 3-item Observation and Interview-based Diurnal Sleepiness Inventory (ODSI) measures the degree of excessive daytime sleepiness in patients with suspected narcolepsy or idiopathic hypersomnia (IH). Second, to assess the correlation between the ODSI and the Epworth Sleepiness Scale (ESS) as well as objective polysomnographic measurements. Third, to test the accuracy of the ODSI to detect narcolepsy or IH (narcolepsy/IH) compared with the ESS.

Methods:

A total of 181 patients complaining of excessive daytime sleepiness filled in the ESS and the ODSI and underwent measurements including actigraphy, full-night polysomnography, Multiple Sleep Latency Test, and 24-hour bedrest sleep recording.

Results:

Narcolepsy or IH was diagnosed in 76 patients. The ODSI found excessive daytime sleepiness in 92.3% of all patients and in 98.7% of those diagnosed with narcolepsy/IH. In the whole population, the ODSI was significantly positively correlated with the ESS (R = .547; 95% confidence interval: .436, .642; P < .001) and weakly with 24-hour total sleep time on bedrest recording (R = .208; 95% confidence interval: .056, .350; P = .047) but not with the Multiple Sleep Latency Test. The ODSI offered a higher negative (92.9%) and positive (44.9%) predictive value to detect narcolepsy/IH than did the ESS (66.7% and 43.3%, respectively). In the IH group, the ODSI’s third-item score (daily sleepiness duration) was significantly higher in patients with than without increased 24-hour total sleep time (P = .023).

Conclusions:

The ODSI is a brief, simple first-line questionnaire that explores both intensity and duration of daytime sleepiness and offers a high sensitivity to detect narcolepsy and IH.

Citation:

Peter-Derex L, Subtil F, Lemaitre G, et al. Observation and Interview-based Diurnal Sleepiness Inventory for measurement of sleepiness in patients referred for narcolepsy or idiopathic hypersomnia. J Clin Sleep Med. 2020;16(9):1507–1515.

BRIEF SUMMARY

Current Knowledge/Study Rationale: Excessive daytime sleepiness is a major public health issue with many causes, including neurological hypersomnia such as narcolepsy and idiopathic hypersomnia. As narcolepsy and idiopathic hypersomnia are infrequent severe conditions whose diagnosis requires highly specialized in-laboratory explorations, accurate tools for excessive daytime sleepiness screening and assessment are needed.

Study Impact: The ODSI’s (Observation and Interview-based Diurnal Sleepiness Inventory’s) high sensibility and simple administration, as well as its ability to quantify sleepiness duration, makes it a good first-line screening tool for narcolepsy and idiopathic hypersomnia, especially before scheduling prolonged bedrest recordings. The weak correlation between self-reported scores and objective explorations observed in our study underlines the complex nature of drowsiness, and highlights the importance of a multimodal comprehensive diagnostic approach.

INTRODUCTION

Excessive daytime sleepiness (EDS) is a major public health issue affecting more than 5% of the general population.¹ It is associated with severe academic and professional issues, cognitive impairments, and increased accidental risk, especially at wheel.^2–4 Its main causes are sleep deprivation, sedative treatments, and sleep disorders, including obstructive sleep apnea (OSA) syndrome, or rare but highly disabling diseases such as narcolepsy and idiopathic hypersomnia (IH; narcolepsy/IH).^1,5

Diagnosis of narcolepsy/IH relies on patient-reported clinical information, self-reported assessments of EDS, and several polysomnographic objective measures.⁶ Questionnaires, such as the Epworth Sleepiness Scale (ESS),⁷ Karolinska Sleepiness Scale,⁸ and Stanford Sleepiness Scale,⁹ are easy to use, inexpensive, but obviously self-reported and associated with confounding factors such as depression.^10,11 Thus, objective laboratory vigilance tests have been proposed: the Multiple Sleep Latency Test (MSLT), which evaluates the propensity to fall asleep when required during 4 or 5 successive nap opportunities, and the Maintenance of Wakefulness Test, which measures the ability not to fall asleep in passive conditions.^12–14 However, such objective tests are time consuming, expansive, and they have been criticized for their “artificial” aspect and their poor correlation with self-reported tests.^13–16 Type 1 narcolepsy benefits from the dosage of hypocretin in the cerebrospinal fluid, but this exploration remains invasive and there is no biological marker available for type 2 narcolepsy and IH.^17,18 As a result, no objective or self-reported method alone is perfect for diagnosing most of neurological hypersomnia.

Thus, the development of new tools for measurement of vigilance may be useful and likely to find application in clinical practice, for screening of patients who require specific in-laboratory explorations, as well as to help provide a better understanding of the multiple dimensions of drowsiness. Recently, the Observation and Interview-based Diurnal Sleepiness Inventory (ODSI) questionnaire has been proposed as a tool for assessing EDS regardless of education and socioeconomic status.^19,20 This new 3-item questionnaire evaluates drowsiness in a passive and in an active situation and quantifies the duration of daytime sleepiness; it can be self-administered or completed by a health professional as an interview with patients and informants. The ODSI allows assessment of sleep propensity and daytime sleepiness duration in an individual’s everyday life without cultural, scholarly, or professional references. Among the numerous factors that may cause a person to be excessively sleepy all the time, the ODSI does not aim to distinguish what those factors may be, but rather distinguishes that excessive sleepiness may be a health concern that warrants further diagnosis. This questionnaire is brief, easy to administer, and exhibits a good internal consistency, test-retest reliability, and concurrent (as compared with the ESS) validity in measuring sleepiness in a population of elderly apneic individuals.^19,20

The purpose of this study investigating the usefulness of the ODSI in a wide population of patients reporting EDS and referred for suspicion of IH or narcolepsy is 3-fold: first, to determine whether the 3-item ODSI questionnaire measures the degree of EDS in patients with suspected narcolepsy/IH; second, to assess the correlation of the ODSI with the ESS and objective polysomnographic measurements; and third, to test the accuracy of the ODSI to detect narcolepsy/IH as compared with the ESS.

METHODS

Patients

Adult patients referred to the Center for Sleep Medicine and Respiratory Diseases with a suspicion of narcolepsy or IH between June 2017 and January 2019 were retrospectively screened to be included in the study. Most of them had been referred by a general practitioner or a pneumologist as they complained of EDS without an obvious etiology according to previous evaluation (sleep diary, ambulatory polygraphy, etc). Thus, full exploration had been performed in our center, including clinical evaluation, questionnaires, actigraphy (for sleep–wake schedules assessment in order to rule out sleep deprivation or sleep–wake rhythm disorder),²¹ and 48-hour in-hospital recording including 1-night polysomnography (PSG; to exclude sleep fragmentation and OSA despite negative polygraphy),²² MSLT, and 24-hour ad libitum sleep monitoring (for diagnosis of narcolepsy and IH).^23,24 In some patients, additional evaluations had been performed: forced awakening test (for sleep inertia assessment),²⁵ psychiatric evaluation (when psychiatric hypersomnia was suspected), and urinary 6-sulfatoxymelatonin assay (in case of sleep–wake rhythm disorder suspicion).²⁶ Patients were classified into 4 diagnostic groups according to clinical and electrophysiological evaluation results, International Classification of Sleep Disorders, Third Edition, criteria,²⁷ and pluri-disciplinary consensus meeting of sleep experts when diagnosis was equivocal, as follows: (1) central disorder of hypersomnolence due to narcolepsy type 1, narcolepsy type 2, or IH (narcolepsy/IH); (2) central disorder of hypersomnolence due to insufficient sleep syndrome; (3) central disorder of hypersomnolence due to other causes (neurological disorder, Kleine Levin syndrome, psychiatric hypersomnia, sedative treatments, sleep–wake rhythm disorder); and (4) hypersomnia secondary to OSA (an apnea-hypopnea index >15 events/h with significant sleep fragmentation [arousal index >20 events/h] was considered likely to explain EDS). The aim of this classification was to distinguish patients with primary neurological hypersomnia in whom vigilance tests or bedrest were needed for diagnosis (narcolepsy or IH) and other patients with either a well-defined diagnosis (OSA diagnosed with PSG, insufficient sleep syndrome diagnosed with actigraphy) or less well-defined diagnosis (the “other” group, which also included rare diseases or “elimination” diagnosis).

All patients had given informed consent for the use of their data for research purposes.

Study protocol and data collection

Clinical data

The following clinical data were reviewed in patients’ medical file: age, sex, medical history, current treatments, body mass index, wake-promoting substances (eg, coffee, cola) consumption.

Questionnaires

As part of routine diagnosis evaluation, patients were asked to complete several questionnaires: ESS,⁷ Pichot fatigue scale,²⁸ Beck Depression Inventory,²⁹ and ODSI, which had been implemented in our Sleep unit since the publications by Onen and colleagues.^19,20 The ODSI is a 3-item assessment tool. The first item examines sleepiness during basic activities of daily living (such as driving, eating, speaking). The second item is related to falling asleep during periods of inactivity (such as reading or watching television). The third item asks about cumulated hours of daytime sleep (including sleepiness, falling asleep, and naps). A weighting of the 3 subscores allows taking into account the fact that falling asleep during active situations (the first item) is likely more dangerous and more abnormal. The total score ranges from 0 (no somnolence) to 24 (excessive somnolence). Previous studies have reported that a cutpoint of 6 was effective for identifying older adults with excessive levels of daytime sleepiness.

Actigraphy

A wrist-worm accelerometer (MotionWatch 8, CamNtech Ltd, Cambridge, UK) was used for ambulatory actigraphy during the 7–14 days preceding hospitalization. Patients completed a sleep diary during the actigraphy recording. MotionWare software was used for data analyses.

PSG

Full-night PSG was conducted in the Sleep Medicine and Respiratory Disease Center. Patients arrived in the late afternoon, returned the wrist accelerometer, and underwent instrumentation for the electrodes and sensors required for PSG. The following signals were recorded: electroencephalogram (Fp2–C4, C4–T4, T4–O2, Fp1–C3, C3–T3, T4–O2, Fz–Cz, Cz–Pz), electro-oculogram, chin and tibialis electromyogram, electrocardiogram, nasal airflow (nasal pressure and thermistor), pulse oxymetry, microphone, and respiratory efforts (thoracic and abdominal). Bedtime was free, but patients were informed that they would be awakened at 7:00 am in the morning for MSLT protocol and that they should have had at least 6 hours of sleep before MSLT.

MSLT

The day following PSG recording, a standard MSLT protocol was administered to patients according to the American Academy of Sleep Medicine guidelines.¹⁴ The initial nap opportunity began 1.5 to 3 hours after termination of the nocturnal recording. Nap opportunities were at 2-hour intervals. Mean sleep latency over all tests and total number of naps containing sleep-onset rapid eye movement (REM) periods were calculated for each patient.

Usually, mean sleep latency of 10 minutes or less indicates objective daytime sleepiness. The current cutoff for the central disorders of hypersomnolence is a mean sleep latency of 8 minutes or less. This cutoff is used as a diagnostic criterion mainly for narcolepsy and IH. In addition, 2 or more sleep-onset REM periods were used to distinguish narcolepsy from IH.²⁷

Twenty-four-hour ad libitum sleep monitoring

After the end of the last nap of the MSLT session, a continuous 24-hour ad libitum sleep monitoring was performed: electroencephalogram, electro-oculogram, chin electromyogram, and electrocardiogram were recorded with an ambulatory device that allowed the patients either to stay in the hospital room or to have a walk around the hospital but not to go back home (in order to prevent them from restricting their sleep time because of everyday life activities). Patients were advised not to resist sleep and to sleep as long as they wished. They were not disturbed during 24 hours and could ask for meals whenever they wanted. This recording allowed the measurement of total sleep time per 24 hours in the absence of constraints. Polysomnograms and MSLT were scored by trained sleep medicine physicians (L.P.-D., H.B., A.B., and F.R.). The 24-hour ad libitum sleep monitoring was performed as a total sleep time of 660 minutes or more as an additional criterion for IH if other polysomnography/MSLT findings are absent.²⁷

Statistical analysis

Quantitative variables were described by the mean (SD) and median (first and third quartiles); qualitative variables were described by the frequency and percentage of each modality (excluding missing data from percentages). Comparisons of results between groups (diagnostic groups, included/excluded patients, groups stratified by age, sex, or body mass index) were performed using the Kruskal-Wallis or Wilcoxon tests for quantitative results, and the chi-square or Fisher’s tests for qualitative results. When the difference was significant, post hoc comparisons were performed to compare groups between one another using the Holm method to take into account multiple testing. Correlations between different tests (ODSI, ESS, MSLT, bedrest) were assessed by the Spearman rank correlation coefficient (R), with its 95% confidence interval (95% CI); the Holm method was used to adjust P values when necessary. The diagnostic value of the tests was assessed by the area under the receiver operating characteristic curve (AUC) with its 95% CI. For sensitivity and specificity, positive- and negative-predictive values were calculated for different thresholds. P < .05 was considered significant. The analyses were performed using the R software version 3.4.4 (R Foundation for Statistical Computing, Vienna, Austria).

RESULTS

Population

A total of 235 consecutive adult patients referred for suspicion of narcolepsy/IH and who had benefited from clinical evaluation, actigraphy, and 48-hour sleep recording were retrospectively identified in our database. Fifty-four patients were excluded because of missing data regarding questionnaires that had not been exhaustively completed by some patients: among them, 34 did not complete any of the 4 questionnaires and at least 1 questionnaire was missing for the 20 others. Among the 181 included patients (mean ± SD age: 38.27 ± 15.93 years; 121 women), 76 (42%) had a final diagnosis of narcolepsy or IH (among which, 64.5% were IH), 24 had a diagnosis of hypersomnia secondary to insufficient sleep syndrome (ISS), 36 had a diagnosis of OSA, and 45 had a diagnosis of “other” hypersomnia including 21 patients with psychiatric hypersomnia. In the narcolepsy/IH group, 17 patients were taking wake-promoting drugs, which had been stopped for at least 4 days before the recordings. A total of 26 patients were taking antidepressant drugs (7 in the narcolepsy/IH group, 2 in the ISS group, 11 in the OSA group, and 6 in the “other hypersomnia” group). Demographic, clinical, and questionnaires results of the patients according to the hypersomnolence cause, are presented in Table 1. The 4 groups of patients significantly differed in age, sex, and body mass index. Post hoc analysis revealed that patients with OSA were significantly more often male (vs narcolepsy/IH, adjusted P = .017; vs ISS, P = .017; vs “others,” P = .003), older (vs narcolepsy/IH, P < .001; vs ISS, P = .011; vs “others,” P = .023), and they had a higher body mass index (P < .001 vs narcolepsy/IH, ISS, and “others”) than patients in the 3 other groups, which did not differ in these variables between one another. The comparison of demographic variables between patients who were (n = 181) and were not (n = 54) included in the study is provided in Table S1. The sex ratio differed between the 2 groups (fewer males in the included group: 33.1% vs 55.4%; P = .005) and the MSLT mean latency was lower in the excluded group (median [quartile 1–quartile 3]: 7.88 [4.81–11.50] vs 10.00 [7.00–13.88], P = .013). No difference was found for the other variables including the distribution of the 4 diagnoses.

Table 1.

Demographic and clinical characteristics of the patients.

	Diagnosis				P
	Narcolepsy/IH (n = 76)	Insufficient Sleep Syndrome (n = 24)	OSA (n = 36)	Others (n = 45)
Age, mean (SD), years	32.87 (14.95)	37.71 (15.22)	49.03 (13.27)*	39.09 (15.80)	<.001
Sex, female, n (%)	53 (69.7)	19 (79.2)	14 (38.9)*	35 (77.8)	<.001
BMI, mean (SD), kg/m2	24.34 (5.58)	23.48 (4.43)	28.52 (5.56)*	23.54 (4.48)	<.001
ODSI
Mean (SD)	16.62 (4.47)	14.71 (5.54)	12.83 (6.60)*	14.58 (5.89)	.023
Median	17.50	16.00	14.00	16.00
Q1–Q3	14.75–20.00	11.75–18.25	8.00–18.00	12.00–19.00
Min–Max	4.00–24.00	3.00–22.00	0.00–23.00	2.00–24.00
ODSI ≥6, n (%)	75 (98.7)	22 (91.7)	29 (80.6)*	41 (91.1)
ESS
Mean (SD)	16.49 (4.35)	15.12 (4.84)	13.31 (4.43)*	15.44 (3.86)	.005
Median	17.00	15.00	14.00	16.00
Q1–Q3	14.00–19.00	11.75–18.50	11.00–16.00	14.00–18.00
Min–Max	6.00–24.00	5.00–23.00	3.00–22.00	1.00–22.00
ESS ≥11, n (%)	68 (89.5)	20 (83.3)	28 (77.8)*	41 (91.1)
Pichot scale
Mean (SD)	19.41 (6.06)	17.71 (8.34)	17.50 (6.81)	21.33 (6.42)	.054
Median	20.50	18.00	17.00	22.00
Q1–Q3	15.00–23.00	12.00–23.25	14.00–22.50	16.00–26.00
Min–Max	2.00–32.00	3.00–32.00	1.00–28.00	5.00–32.00
BDI
Mean (SD)	9.84 (6.94)	10.22 (7.57)	10.56 (6.85)	13.36 (8.29)	.105
Median	9.00	7.00	10.00	12.00
Q1–Q3	5.00–13.00	4.50–14.50	5.00–16.25	6.00–19.00
Min–Max	0.00–31.00	1.00–29.00	0.00–22.00	0.00–37.00

BDI = Beck Depression Inventory; BMI = body mass index; ESS = Epworth Sleepiness Scale; IH = idiopathic hypersomnia; Max = maximum; Min = minimum; ODSI = Observation and Interview-based Diurnal Sleepiness Inventory; OSA = obstructive sleep apnea syndrome; Q, quartile. Post hoc analyses: *Significant (P < .05) differences between groups. Patients with OSA were significantly older, less often female, and more obese than the 3 other groups; they had significantly lower ODSI and ESS scores than patients with narcolepsy/IH.

Results of questionnaires and sleep recordings

With regard to questionnaires, 92.3% and 86.7% of patients had abnormal ODSI and ESS scores, respectively. No difference between groups was seen for Pichot (despite a strong trend, P = .054, with higher values observed in the “other” group) and Beck Depression Inventory scales, nor for the proportion of patients with abnormal ESS scores. After adjusting for sex, which was found to be a determining factor of Pichot scale (see Table S2), the 4 groups differed (P = .04516), with no significant post hoc results. ESS mean score, ODSI mean score, and proportion of patients with abnormal ODSI score were significantly higher in the narcolepsy/IH group than in the OSA group (respectively, adjusted P = .003, P = 0.24, P = .003). All patients but one had an abnormal ODSI score in the narcolepsy/IH group.

Detailed results of sleep recordings are provided in Table S3. Mean MSLT latency was abnormal in 61.8% of patients with narcolepsy/IH, but also in 8.3% of patients with ISS, 14.7% of patients with OSA, and 15.6% in the “other” group. Five out of the 29 recorded sleep-onset REM periods were found in patients without narcolepsy/IH. As expected, sleep was more disturbed in patients with OSA with reduced sleep efficiency as compared with the narcolepsy/IH group (P = .014) and they had an increased arousal index as compared with the 3 other groups (P < .001).

Correlations between questionnaires and objective tests

Results of correlation analyses are presented in Figure 1. The ODSI and ESS were significantly positively correlated with each other in the whole population (R = .547; 95% CI: .436, .642; P < .001) and in the narcolepsy/IH group (R = .562; 95% CI: .386, .699; P < .001). A weak negative correlation was found between the ESS and MSLT both in the whole population (R = −.208; 95% CI: −.344, −.063; P = .021) and in patients with narcolepsy/IH (R = −.304; 95% CI: −.495, −.084; P = .023). A weak negative correlation was also found between the ODSI and MSLT in the whole population (R = −.175; 95% CI: −.314, −.029; P = .038), whereas this correlation did not remain significant after correction for multiple correlations in patients with narcolepsy/IH (R = −.158; 95% CI: −.370, .070; P = .173).

Figure 1. Correlations between self-reported and objective tools.

The ODSI and ESS are significantly positively correlated with each other in the whole population (R = .547; 95% CI: .436, .642; P < .001) (A) and in patients with narcolepsy/IH (R = .562; 95% CI: .386, .699; P < .001) (B). In the whole population, the MSLT mean latency is significantly but weakly negatively correlated with the ESS (R = −.208, 95% CI: −.344, −.063; P = .021) (C) and the ODSI is significantly but weakly positively correlated with the total sleep time assessed by 24-hour Holter recording (R = .208; 95% CI: .056, .350; P = .047) (D). P values are adjusted for multiple comparisons. CI = confidence interval; ESS = Epworth Sleepiness Scale; IH = idiopathic hypersomnia; MSLT = Multiple Sleep Latency Test; ODSI = Observation and Interview-based Diurnal Sleepiness Inventory.

A weak but significant positive correlation was found between the ODSI and total sleep time recorded during 24-hour bedrest in the whole population (R = .208; 95% CI: .056, .350; P = .047) but not in the narcolepsy/IH group (R = .279; 95% CI: .036, .491; P = .127), or between the ESS and total sleep time recorded during 24-hour bedrest in the whole population (R = .117; 95% CI: −.037, .266; P = .407) and in the narcolepsy/IH group (R = .070; 95% CI: −.179, .311; adjusted P = .685, P = 1). No significant correlation was found either between MSLT and 24-hour bedrest results in the whole population (R = −.136; 95% CI: −.285, .019; P = .342) or in the narcolepsy/IH group (R = .121; 95% CI: −.129, .356; P = .685).

Accuracy of the ODSI for the detection of narcolepsy/IH

Global ODSI score

Receiver operating characteristic curves of diagnosis value for the ODSI and the ESS are presented in Figure 2: AUCs were, respectively, .622 (with 95% CI: .541, .704) and .623 (with 95% CI: .539, .706), demonstrating very close diagnostic accuracies of the 2 questionnaires.

Figure 2. Accuracy of self-reported tools alone to detect narcolepsy/IH.

ROC curves of the ODSI and ESS; both questionnaires exhibit a high sensitivity. ESS = Epworth Sleepiness Scale; IH = idiopathic hypersomnia; ODSI = Observation and Interview-based Diurnal Sleepiness Inventory; ROC = receiver operating characteristic.

Sensitivity, specificity, and positive- and negative-predictive value of ESS and ODSI, according to different thresholds for narcolepsy/IH diagnosis, are presented in Table 2. Among patients with narcolepsy/IH, 98.7% and 89.5% had abnormal scores on the ODSI (score ≥6) and the ESS (score ≥11), respectively. As a result, the ESS, and especially the ODSI, had an excellent sensitivity; ODSI with a threshold of 6 or higher exhibited the highest negative-predictive value (.929; 95% CI: .661, .998). On the other hand, positive-predictive values were low (.449). Only a high (≥18) threshold for the ODSI allowed reaching a satisfactory specificity (.676; 95% CI: .578, .764).

Table 2.

Diagnostic value of the ODSI and ESS in the narcolepsy/IH group without any additional criteria.

Test Score	Sensitivity [95% CI]	Specificity [95% CI]	Positive-Predictive Value [95% CI]	Negative-Predictive Value [95% CI]
ODSI ≥6	.987 [.929, 1.000]	.124 [.068, .202]	.449 [.372, .528]	.929 [.661, .998]
ODSI ≥9	.921 [.836, .970]	.229 [.152, .321]	.464 [.382, .546]	.800 [.614, .923]
ODSI ≥13	.829 [.725, .906]	.352 [.262, .452]	.481 [.393, .570]	.740 [.597, .854]
ODSI ≥18	.500 [.383, .617]	.676 [.578, .764]	.528 [.407, .647]	.651 [.554, .740]
ESS ≥11	.895 [.803, .953]	.152 [.090, .236]	.433 [.354, .514]	.667 [.447, .844]

Sensitivity, specificity, and positive- and negative-predictive values are presented for different ODSI cutpoints and for the ESS. CI = confidence interval; ESS = Epworth Sleepiness Scale; IH = idiopathic hypersomnia; ODSI = Observation and Interview-based Diurnal Sleepiness Inventory.

Subscore analysis

A more detailed analysis of the ODSI results was performed, using subscores on the 3 items. Results for each item are detailed in Table 3. Patients with OSA had significantly lower scores on item 2 than patients with narcolepsy/IH (P < .001). ODSI score did not differ between patients with narcolepsy and IH. However, subscores exhibited differences for item 1 and 2 (significantly higher in the narcolepsy group; P = .028 and P = .018, respectively) but no significant difference was found for item 3 (P = .834). In order to examine if item 3 (which aims to quantify sleepiness duration) was associated with a specific phenotype of IH, we compared the subgroup of IH with increased total sleep time on bedrest (24-hour total sleep time ≥660 minutes, n = 30) with the subgroup of IH without long sleep duration (n = 19); item 3 was significantly higher in the former group (3.03 ± 1.54 vs 1.80 ± 1.42, P = .023).

Table 3.

The ODSI subscores according to diagnosis.

	Diagnosis						Total (n = 181)
	Narcolepsy/IH			Insufficient Sleep Syndrome (n = 24)	OSA (n = 36)	Others (n = 45)
	All Patients (n = 76)	Narcolepsy (n = 27)	Idiopathic Hypersomnia (n = 49)
ODSI
Mean (SD)	16.62 (4.47)	17.85 (4.19)	15.94 (4.51)	14.71 (5.54)	12.83 (6.60)	14.58 (5.89)	15.10 (5.59)
Median	17.50	19.00	17.00	16.00	14.00	16.00	16.00
Q1–Q3	14.75–20.00	15.50–20.00	14.00–18.00	11.75–18.25	8.00–18.00	12.00–19.00	12.00–19.00
Min–Max	4.00–24.00	4.00–24.00	6.00–24.00	3.00–22.00	0.00–23.00	2.00–24.00	0.00–24.00
ODSI item 1
Mean (SD)	8.53 (3.68)	9.56 (3.14)	7.96 (3.85)*	7.83 (4.35)	6.75 (4.72)	7.16 (4.37)	7.74 (4.19)
Median	10.00	10.00	9.00	10.00	8.00	8.00	9.00
Q1–Q3	8.00–11.00	9.00–12.00	7.00–10.00	7.00– 11.00	0.00–10.00	7.00–10.00	7.00–11.00
Min–Max	0.00–12.00	0.00–12.00	00.00–12.00	0.00– 12.00	0.00–12.00	0.00–12.00	0.00–12.00
ODSI item 2
Mean (SD)	5.41 (0.88)	5.70 (0.67)	5.24 (0.95)*	5.00 (1.14)	4.19 (1.72)**	5.11 (1.25)	5.04 (1.28)
Median	6.00	6.00	6.00	5.00	4.50	6.00	6.00
Q1–Q3	5.00–6.00	6.00–6.00	6.00–6.00	4.00– 6.00	3.75–6.00	4.00–6.00	4.00–6.00
Min–Max	2.00–6.00	3.00–6.00	2.00–6.00	2.00–6.00	0.00–6.00	1.00–6.00	0.00–6.00
ODSI item 3
Mean (SD)	2.71 (1.63)	2.67 (1.54)	2.73 (1.69)	1.88 (1.54)	1.92 (1.71)	2.27 (1.72)	2.33 (1.68)
Median	2.50	2.00	3.00	2.00	2.00	2.00	2.00
Q1–Q3	1.75–4.00	1.00–4.00	2.00–4.00	0.75–3.00	0.00–3.00	1.00–3.00	1.00–3.00
Min–Max	0.00–6.00	1.00–6.00	0.00–6.00	0.00–5.00	0.00–6.00	0.00–6.00	0.00–6.00

IH = idiopathic hypersomnia; Max = maximum; Min = minimum; ODSI = Observation and Interview-based Diurnal Sleepiness Inventory; OSA = obstructive sleep apnea syndrome; Q, quartile. *Patients with narcolepsy have significantly higher scores for ODSI item 1 and 2 than patients with IH (P = .028 and P = .018, respectively). **Patients with OSA have lower scores for ODSI item 2 than patients with narcolepsy/IH (P < .001).

DISCUSSION

In the present study we assessed the diagnostic accuracy of a 3-item questionnaire, the ODSI, in a wide population of middle-aged adult patients complaining of EDS due to different causes. The ODSI had been previously evaluated in older adults with OSA.¹⁹ We report here that, in younger patients with OSA, but also central disorder of hypersomnolence including narcolepsy and IH, ISS, and psychiatric hypersomnia, the ODSI is highly correlated with EDS complaints and with the ESS scores. In the specific population of patients diagnosed with narcolepsy and IH, its negative-predictive value is higher than that of the ESS, and subitem analysis allows discrimination between narcolepsy and IH, as well as between hypersomnia with and without long duration. However, the ODSI specificity remains low and, like the ESS, the ODSI is poorly correlated with objective vigilance tests.

Early diagnosis of narcolepsy/IH is warranted, given the clinical, social, and economic burden of these diseases.^30,31 Thus, high-sensitivity, simple, first-step screening tools that could be used in the primary care setting are needed to identify patients with hypersomnolence who require sleep exploration. The ODSI is an easy-to-use and brief questionnaire, that can be either self-, proxy- or physician-administered.^19,20 It is usable by nonspecialist physicians and may apply to a wide population of adult patients without social or occupational reference, whereas patients may not engage in all of the activities identified by the ESS (eg, some patients do not drive or barely read).³² The ODSI reaches a high sensitivity (98.7%) and negative-predictive (92.9%) value that allow considering it as a first-line screening questionnaire. Moreover, the ODSI explores different components of hypersomnolence, including sleep propensity (sleepiness) but also sleep duration, which may be particularly interesting in the setting of IH in which some patients do not complain of EDS but rather experience an increase in sleep duration and sleep inertia.^23,33 Interestingly, item 3 (“estimated sleepiness duration”) of the ODSI score was found to discriminate patients with hypersomnia with and without increase in sleep duration, as assessed by 24-hour Holter recording. The ODSI may thus be used to identify patients with IH who may need prolonged 24- or 36-hour bedrest because of a “long sleep time” phenotype.^23,33,34

Accurate screening of patients who require evaluation in specialized centers is also needed in order to avoid unnecessary, time-consuming, and costly evaluations.^5,35–38 Diagnosis of narcolepsy/IH was confirmed in less than half of the patients admitted to our center with suspected diagnosis of narcolepsy or IH. This emphasizes both the fact that numerous other causes lead to EDS and numerous patients remain overreferred for narcolepsy/IH, especially patients with polygraphy-negative OSA or patients with chronic sleep deprivation, in whom a sleep diary may not be reliable and actigraphy is not always performed despite recommendations.^21,39,40 According to the International Classification of Sleep Disorders, Third Edition, PSG followed by MSLT and/or 24- to 36-hour bedrest are considered the gold-standard evaluations for narcolepsy/IH diagnosis.²⁷ However, abnormal short sleep latency and sleep-onset REM periods, as well as increased total sleep duration can be observed in other conditions.⁴¹ In our population, 14 patients without narcolepsy/IH had abnormal MSLT. On the other hand, sensitivity of this test depends on normative criteria but is not perfect.⁴² This stresses the fact that objective measurements cannot be used in isolation to confirm or exclude narcolepsy/IH in patients with EDS and that they need to be interpreted in conjunction with the comprehensive analysis of clinical evaluation and questionnaires. Moreover, the weak correlation between self-reported scores and objective explorations observed in our study and by other teams underlines the complex nature of drowsiness, and highlights the importance of a multimodal comprehensive diagnostic approach.^13,43–45

We acknowledge several limitations of our work. First, this was a retrospective study; 54 out of our 235 patients were excluded because of missing data in questionnaires. The comparison between included and excluded patients revealed that the 2 groups differed in sex ratio and MSLT mean latency. More females were present in the included group, potentially because sex may influence survey response rates in general.⁴⁶ However, since sex was not a determining factor of the ODSI in our study, we do not believe that this bias could have influenced our results. The lower MSLT latency in the excluded group may suggest that these patients experienced a more severe sleepiness that had prevented them from reading and filling in questionnaires. This may limit the generalization of our results to more severe populations. Second, our narcolepsy/IH population does not reflect the epidemiology of narcolepsy and IH, even if the latter is less well known.^47–49 Indeed, patients with IH were twice as prevalent, and NT2 patients were as numerous as NT1 patients. This may be explained by the fact that, when patients present with a suspicion of narcolepsy with highly suggestive symptoms, they benefit from PSG followed by MSLT only. The exhaustive evaluation including actigraphy, PSG, MSLT, and 24-hour bedrest is reserved for patients with a suspicion of IH or without a clear diagnosis orientation. Third, we decided to group different diagnoses (eg, hypersomnia due to substances, to psychiatric or medical conditions, sleep–wake rhythm disorder) into the same category in order not to increase the number of subgroups and as we wanted to study the most frequent causes of EDS (OSA and ISS) as well as the main group of our study, narcolepsy/IH. It is also important to note that the ODSI has not yet been specifically validated in a population other than elderly individuals with OSA. Finally, the gold standard for diagnosis of many causes of hypersomnolence is questionable; diagnostic criteria are evolving, and some are not clearly defined (such as ISS, which is a relative notion). We used International Classification of Sleep Disorders, Third Edition, criteria, but also, in many cases, pluri-disciplinary consensus of experts including neurologists, psychiatrists, and pneumologists, all of whom are specialized in sleep medicine.

CONCLUSIONS

In conclusion, the ODSI is a simple, easy-to-use, questionnaire correlated with the ESS in a population of young and middle-aged adults with hypersomnolence due to various sleep pathologies. The ODSI scores separately perceived duration of sleepiness, sleepiness in conditions where the individual should be active, and sleepiness in resting or passive conditions. In the context of narcolepsy/IH, the ODSI has a high sensitivity and a good negative-predictive value, which makes it a good screening test and a first step before scheduling further sleep evaluation, especially for patients with IH in whom prolonged bedrest recording is required for diagnosis.

DISCLOSURE STATEMENT

All authors have seen and approved the manuscript. The authors report no conflicts of interest.

ABBREVIATIONS

CI

confidence interval

EDS

excessive daytime sleepiness

ESS

Epworth Sleepiness Scale

ISS

insufficient sleep syndrome

IH

idiopathic hypersomnia

MSLT

Multiple Sleep Latency Test

ODSI

Observation and Interview-based Diurnal Sleepiness Inventory

OSA

obstructive sleep apnea

PSG

polysomnography

REM

rapid eye movement