Keywords
Disorders of Excessive Somnolence - Sleepiness - Epworth Sleepiness Scale - Students,
Medical
INTRODUCTION
Excessive daytime sleepiness (EDS) has been described as a symptom that arises at
any time and refers to the propensity to get drowsy or even fall asleep when the intention
and expectation would be to stay awake.[1] It is associated with difficulty in maintaining attention and impairment in reasoning
and memory in addition to affecting the quality of life and academic and professional
performance.[2]
Several methods and tools have been developed to assess EDS, ranging from detailed
clinical investigation to specific tests such as polysomnography, wakefulness maintenance
test, and multiple sleep latency testing.[3] There are also subjective tests, such as scales that evaluate EDS through the individual's
self-report, among which the Epworth Sleepiness Scale (ESS) can be cited.[1]
The ESS was originally developed in English to evaluate EDS in adult populations;
however, it has also been successfully used among children and adolescents.[4] It has been translated into several languages and used in different cultures,[5]
[6]
[7] including the Brazilian population.[8] It has been used both in populations in general[9] as well as in clinical evaluations.[10]
The ESS, in its original validation study, presented a high level of internal consistency,
with Cronbach's α ranging from 0.88 for individuals with sleep disorders to 0.73 for
medical university students.[11] In the only ESS validation study for the Brazilian population, the scale was tested
in patients aged 18 to 65 years with specific characteristics, including evaluation
for primary snoring, insomnia, and obstructive sleep apnea-hypopnea syndrome compared
with a control group with normal sleep habits and without apparent snoring, which
presented an internal consistency with a Cronbach's α of 0.83.[8] However, in addition to the population being specific, only the first validation
stage was performed: translation and back translation.
Construct validity seeks to test whether the instrument can adequately represent the
theoretical construct that it seeks to measure; thus, one can determine whether the
ESS evaluates EDS in a valid and precise way, according to the principles of psychometrics.[12] While internal consistency refers to the homogeneity of the scale items,[12]
[13] this can indicate whether all ESS items necessarily measure the same characteristic
(EDS).
Taking into account that EDS has a high prevalence among university students,[14]
[15]
[16] and given the impact that it can have on the performance and academic training of
this population,[2] the present study aimed to evaluate the psychometric properties of the ESS, especially
its construct validity and internal consistency in Brazilian medical students.
METHODS
Samples and participants
Two samples were obtained from a cross-sectional academic study conducted in 2018,
including a population of 2,295 health students (1,582 medical students) of the Rio
Verde, Aparecida de Goiânia, and Goianésia campuses of Universidade do Rio Verde –(UniRV),
a public university located in the Brazilian Midwest. The present study included all
university students aged over 18 years enrolled in health courses across the three
campuses of both sexes. Students who reported the use of medications for sleep, such
as sedatives, tranquilizers, or anxiolytics, as well as pregnant or breastfeeding
students, were excluded, as these conditions may have affected the outcomes.[16]
For the present study, from a total of 1,582 medical students, 800 were selected to
form 2 independent samples of 400 students each. This included a randomized sample
of 400 students (200 of each sex) for Exploratory Factor Analysis (EFA) and another
randomized sample of 400 students (200 of each sex) for Confirmatory Factor Analysis
(CFA). This sample size of 400 students in each group was determined to obtain factor
loadings of 0.30 or higher as statistically significant,[17] considering a statistical power of 80% and a confidence level of 95%.
The current study was submitted and approved by the Research Ethics Committees of
Universidade do Vale do Rio dos Sinos (process no. 2.892.764) and UniRV (process no.
2.905.704). All students who agreed to participate in the study signed an Informed
Consent Form before answering the questionnaires, according to the ethical aspects
for research involving human beings arranged in resolution 466/2012 of the National
Health Council of Brazil. All the student data were maintained with confidentiality
and compliance with the Declaration of Helsinki. This study was reviewed by two Research
Ethics Committees; however, data collection was conducted exclusively at UniRV.
Instruments
Excessive daytime sleepiness was assessed with the Portuguese-language version of
the ESS,[1]
[3]
[8] which includes a list of eight situations to mark the probability of napping or
sleeping, and not just feeling tired, in each of them. The eight situations investigated
are: “sitting and reading”; “watching TV”; “sitting inactive in a public place (e.g.,
a theater or a meeting)”; “as a passenger in a car for an hour without a break”; “lying
down to rest in the afternoon when circumstances permit”; “sitting and talking to
someone”; “sitting quietly after a lunch without alcohol” and “in a car, while stopped
for a few minutes in traffic.”
In each of these situations, the student was instructed to respond on a 4-point scale,
whose scores range from 0 to 3, in which 0 - “no chance of dozing,” 1 - “slight chance
of dozing,” 2 - “moderate chance of dozing” and 3 - “high chance of dozing,” represent
the different levels of sleepiness in each of the situations. The student was always
oriented to consider their way of life taken recently and was instructed to imagine
how it would affect them, in case any of these situations had not been encountered.
The score was obtained by applying the ESS ranges from 0 to 24, calculated by adding
the individual scores for each of the 8 items. The interpretation of the score is
divided into four categories: 0 to 7 (“it is unlikely that you are abnormally sleepy”);
8 to 9 (“you have an average amount of daytime sleepiness”); 10 to 15 (“you may be
excessively sleepy depending on the situation, and may want to consider seeking medical
attention”); and 16 to 24 (“you are excessively sleepy and should consider seeking
medical attention”).[1]
[11]
In addition to the ESS, the students answered a standardized and precoded questionnaire
containing sociodemographic variables (age, sex, skin color/race, marital status,
and economic class), to characterize the sample. Economic class was determined using
the Brazilian Association of Research Companies (Associação Brasileira de Empresas
de Pesquisa – ABEP, in Portuguese) scale, which estimates the purchasing power of
individuals and families based on household possessions and the education level of
the head of the household. The classes were subsequently categorized as high (A),
middle (B), and low (C/D/E). The questionnaire administration was conducted in the
classroom, where the instructions were read aloud by the researcher responsible to
facilitate the response, resolve doubts, and minimize loss of data from respondents.
After its completion, the questionnaire was deposited in a sealed urn, and, subsequently,
the codification of the same was performed.
Data analysis
Double data entry was conducted using Epidata (Centers for Disease Control and Prevention,
Atlanta, GA, USA), version 3.1, followed by a subsequent comparison to identify and
rectify potential typos. Socioeconomic and demographic characteristics such as sex,
age, skin color/race, marital status, and economic class were analyzed by means of
absolute and relative frequencies.
Exploratory factor analyses (EFAs) and confirmatory factor analyses (CFAs) for the
8 ESS items were conducted in the Mplus program, version 8.4 (Muthén & Muthén, Los
Angeles, CA, USA),[18] using the weighted least squarest mean and variance (WLSMV) adjusted estimator for
categorical variable scan analysis. Loads with values ≥ 0.30 were considered acceptable;
this is the minimum value required for the variable to be a useful representative
of the factor.[19] For the interpretation of the results, Geomin oblique factorial rotation was used,
because it allows the factors to be correlated with each other.[17]
The CFA was performed to confirm the hypothetical factorial structure found in the
EFA, thus identifying the validity of the ESS construct. To measure the absolute fit
quality indices of the factorial models, we used the Chi-squared (χ2), standardized root mean square residual (SRMR), and root mean square error of approximation
(RMSEA) tests. For the χ2 its value had to accept the null hypothesis, with p-value > 0.05. For the SRMR and RMSEA, values close to or below 0.08 and 0.06 respectively
indicated a satisfactory adjustment of the model. Two other relative adjustment indexes
used were the Tucker-Lewis Index (TLI) and the Comparative Fit Index (CFI), with values
from 0.90 to 0.95 expected as indicative of acceptable adjustment.[20]
The Cronbach's α coefficient (α) was used to evaluate the internal consistency of
the scale. Internal consistency assesses whether the domains of the instrument in
question measure the same characteristic, that is, they measure its homogeneity. The
α coefficient corresponds to an index used to measure the reliability of the internal
consistency type of a scale, that is, to evaluate the magnitude to which the items
of an instrument are correlated.[13] Values of α ≥ 0.7 for each factor show a factorial structure with good internal
consistency.[21]
RESULTS
[Table 1] shows the distribution of samples according to the sociodemographic characteristics
of the university students included in the EFA (n = 400) and the CFA (n = 400). A similar distribution of the evaluated characteristics was observed between
the 2 samples. About ⅓ of the two samples was formed by university students aged between
20.1 and 22 years, 32.3% in the sample used in the EFA and 36.3% in the one for CFA.
Most of the students in both samples, EFA and CFA, respectively, reported having white
skin color (62%; 61%), not having a partner (88.97%; 90.73%), and belonging to economic
class A (54.52%; 57.91%).
Table 1
Distribution of the samples according to the demographic and socioeconomic characteristics
of university students in the Brazilian Midwest (EFA = 400; CFA = 400)
|
Variables
|
Categories
|
EFA Sample
|
CFA Sample
|
|
n
|
%
|
n
|
%
|
|
Sex
|
Female
|
200
|
50.0
|
200
|
50.0
|
|
Male
|
200
|
50.0
|
200
|
50.0
|
|
Age (years)
|
18–20
|
99
|
24.81
|
89
|
22.31
|
|
20.1–22
|
129
|
32.33
|
144
|
36.09
|
|
22.1–24
|
104
|
26.79
|
95
|
23.81
|
|
> 24
|
67
|
16.79
|
71
|
17.79
|
|
Skin color/race
|
White
|
248
|
62.0
|
244
|
61.0
|
|
Brown/Black
|
142
|
35.5
|
143
|
35.75
|
|
Other
|
10
|
2.5
|
13
|
3.25
|
|
Marital status
|
With partner
|
44
|
11.03
|
37
|
9.27
|
|
Without partner
|
355
|
88.97
|
362
|
90.73
|
|
Economic class*
|
Class A (high)
|
211
|
54.52
|
227
|
57.91
|
|
Class B (middle)
|
150
|
38.76
|
136
|
34.69
|
|
Class C, D, E (low)
|
26
|
6.72
|
29
|
7.40
|
Abbreviations: CFA, confirmatory factor analysis; EFA, exploratory factor analysis.
Note: *Economic class was determined using the Brazilian Association of Research Companies
(ABEP) scale, which estimates the purchasing power of individuals and families based
on household possessions and the education level of the head of the household.
The EFA showed a good fit for both the models evaluated (unifactorial and bifactorial),
fixing the extraction in one and two factors, the factor loadings presented values
higher than 0.5 in all items and solutions evaluated. As expected, the two-factor
solution presented a better adjustment (χ[2] = 0.627, SRMR = 0.029, RMSEA = 0.001, CFI = 1, and TLI = 1) ([Table 2]). When evaluating the distribution of factor loadings, it was observed that questions
1, 2, 4, 5, and 7 presented higher factor loadings in factor 1, while questions 3,
6, and 8 had higher factor loadings in factor 2. Thus, factor 1 was named ‘sleepiness
in a state of rest’ and those grouped in factor 2 displayed ‘active sleepiness’ ([Table 2]). In the internal consistency analysis, Cronbach's α coefficients (α) were found
with values higher than 0.7 in all factorial models tested ([Table 2]).
Table 2
Exploratory factor analysis and internal consistency of the Epworth Sleepiness Scale
in university students from the Brazilian Midwest, 2018 (n = 400)
|
Epworth Sleepiness Scale**
|
Total Sample (n = 400)
|
|
Item
|
1 Factor
|
2 Factors
|
|
How likely are you to doze off or fall asleep in the following situations? You should
rate your chances of dozing off, not just feeling tired. Even if you have not done
some of these things recently try to determine how they would have affected you.
|
RV
|
F1
|
RV
|
F1
|
F2
|
|
1) Sitting and reading
|
0.57
|
0.654*
|
0.56
|
0.548*
|
0.170
|
|
2) Watching TV
|
0.65
|
0.589*
|
0.60
|
0.631*
|
0.000
|
|
3) Sitting inactive in a public place (e.g., a theater or a meeting)
|
0.52
|
0.696*
|
0.47
|
0.244*
|
0.555*
|
|
4) As a passenger in a car for an hour without a break
|
0.76
|
0.489*
|
0.76
|
0.338*
|
0.203
|
|
5) Lying down to rest in the afternoon when circumstances permit
|
0.65
|
0.596*
|
0.55
|
0.699*
|
−0.054
|
|
6) Sitting and talking to someone
|
0.59
|
0.640*
|
0.40
|
−0.014
|
0.780*
|
|
7) Sitting quietly after a lunch without alcohol
|
0.52
|
0.691*
|
0.51
|
0.567*
|
0.190
|
|
8) In a car, while stopped for a few minutes in traffic
|
0.66
|
0.583*
|
0.57
|
0.062
|
0.619*
|
|
Internal consistency - Cronbach's Alpha (α)
|
–
|
0.75
|
–
|
0.70
|
0.75
|
|
Eigenvalue
|
–
|
3.643
|
–
|
3.643
|
0.992
|
|
Chi-squared (χ2);
p-value
|
|
40.685 (20);
p = 0.0041
|
|
10.813 (13);
p = 0.6265
|
|
SRMR
|
|
0.062
|
|
0.029
|
|
RMSEA
|
|
0.051
|
|
0.001
|
|
CFI
|
|
0.979
|
|
1.000
|
|
TLI
|
|
0.971
|
|
1.000
|
Abbreviations: CFI, Comparative Fit Index; RMSEA, root mean square error of approximation;
RV, residual variance; SRMR, standardized root mean square residual; TLI, Tucker-Lewis
Index.
Notes: **Model with 1 factor: F1 = general sleepiness; model with 2 factors: F1 = sleepiness
at rest; F2 = drowsiness in activity; commonalities: RV; Kaizer method: > 1/χ2 (p-value > 0.05); Cronbach's α coefficient > 0.7; SRMR < 0.08; RMSEA < 0.06; CFI < 0.9;
and TLI < 0.90.
[Figure 1] shows the CFA developed with a single factor and with 2 factors, identifying better
adjustment measures in the factorial model with 2 factors (SRMR = 0.053, RMSEA = 0.095,
CFI = 0.937, and TLI = 0.908, p = 0.001).
Notes: *f1: Drowsiness in General; sleep 1, sitting and reading; sleep 2, watching
TV; sleep 3, sitting inactive in a public place (such as a theater or a meeting);
sleep 4, as a passenger in a car for an hour without a break; sleep 5, lying down
to rest in the afternoon when circumstances permit; sleep 6, sitting and talking to
someone; sleep 7, sitting quietly after a lunch without alcohol; sleep 8, in a car,
while stopped for a few minutes in traffic.
Figure 1 Confirmatory Factor Analysis of the Epworth Sleepiness Scale (ESS) with 1 factor
and 2 factors among university students from the Brazilian Midwest, 2018 (n = 400).
DISCUSSION
The current study aimed to investigate the construct validity and internal consistency
of the ESS for Brazilian medical students. Our findings showed appropriate psychometric
properties, including an acceptable construct validity and internal consistency in
this population group.
The ESS has presented itself as a solid scale, in psychometric terms, in several studies
conducted around the world, with internal consistency and construct validity considered
appropriate,[7]
[22]
[23] and demonstrating psychometric properties from acceptable to robust,[5] including high sensitivity and specificity in its original[4] as well as in translated versions.[5]
[22]
[23]
Previous literature has shown that the ESS can be used for measuring the propensity
to EDS in university populations of different nationalities and cultures.[6]
[7]
[22]
[23] However, the present study investigates the psychometric properties of the ESS in
Brazilian university students, in whom adequate psychometric properties were observed
in relation to construct validity and internal consistency of ESS.
In the present study, a good internal consistency (α = 0.75) was verified, corroborating
the results of previous studies conducted with university students in Ethiopia (α = 0.75),[23] Peru (α = 0.85), Thailand (α = 0.78),[7] and India (α = 0.86).[22] A systematic review study on the psychometric properties of the ESS showed that
Cronbach's α coefficients tend to range from 0.7 to 0.9, and lower values are generally
found among non-clinical samples, such as students and community samples.[5]
Although the ESS was originally developed and validated as a unifactorial scale,[1]
[24] in our study, the results of exploratory and confirmatory factor analyses revealed
the 2-dimensionality of this scale among Brazilian university students, with the items
related to the evaluation of drowsiness in a state of rest being grouped in factors
1 and 2, and drowsiness evaluated in activity. In this sense, previous studies have
also demonstrated that the eight items of the ESS do not evaluate a one-dimensional
construct, both in clinical populations[5]
[11] and non-clinical ones.[6]
[7]
[22]
[23]
The identification of these two forms of drowsiness reflects the importance of the
different types of situations in which the individual may present drowsiness, with
the situations that allow EDS at rest being perfectly acceptable as a form of rest
and momentary relaxation. However, sleepiness during activity brings great concerns,
since this type is considered socially unacceptable because it is related to a higher
probability of traffic accidents, occupational accidents, and directly interferes
with the health of the individual.[25]
[26]
With regard to this two-dimensionality of factors for the evaluation of sleepiness
in resting and active situations, a study conducted with American and Austrian university
students is in line with our results because they refer to this issue of passive and
active sleepiness.[6] However, it is noteworthy that, unlike our study, the abovementioned study did not
consider question 3 of the ESS (“sitting inactive in a public place setting, such
as in a theater or a meeting”) as a factor that refers to situations of activity.
The inclusion of question 3 in our study's characterization of drowsiness may be justified
by the social phenomenon of fear of inadequacy in public settings, particularly concerning
the inappropriateness of sleeping in public. The need to remain alert in social situations
may induce a heightened state of vigilance that counteracts the sensation of fatigue,
potentially increasing anxiety and, consequently, furthering vigilance, which can
affect the perception of drowsiness.[27] Thus, even in situations in which the body is at rest, such as during a meeting
or theater performance, the brain may remain in a heightened state of alertness in
response to concerns about social behavior and the expectation of attentiveness.
The strengths of the present study include a relatively large sample size that was
sufficient to perform exploratory and confirmatory factor analyses, including homogeneity
between the two samples of university students explored. We highlight a standard administration
of the instruments used in the study and methodological rigor in data collection.
However, it is noteworthy that the present study was limited to evaluating the internal
and construct validity of the ESS scale, not contemplating other forms of validity.
Moreover, the modern era, marked by the pervasive use of smartphones and tablets,
has substantially transformed sleep and wakefulness patterns, emphasizing the need
to reassess the ESS. Integrating questions that address the impact of blue light exposure
and electronic device use before bedtime could provide a more comprehensive assessment
of sleepiness in contemporary life. Evidence suggests that interaction with digital
technology not only affects sleep quality but also influences daytime sleepiness,[28] while digital content consumption may lead to visual overstimulation and circadian
rhythm disruption.[29] Therefore, revising the scale to incorporate these modern influences is essential
to understand how technological factors shape sleepiness in contemporary society.
In conclusion, the present study showed appropriate psychometric properties for the
ESS in medical students, including an acceptable construct validity and internal consistency.
Thus, the ESS may be considered as suitable to assess EDS in university students,
especially medical students. A two-dimensional factor structure was identified, including
factors for assessing drowsiness at rest and drowsiness in activity. However, further
investigations on this factorial structure of the ESS need to be developed, since
presenting EDS in activity is unusual and deserves to be explored.
Bibliographical Record
Renato Canevari Dutra da Silva, Anderson Garcez, Adriana Vieira Macedo Brugnoli, Marcos
Pascoal Pattussi, Maria Teresa Anselmo Olinto. Psychometric properties of the Epworth
Sleepiness Scale in Brazilian medical students. Arq Neuropsiquiatr 2025; 83: s00451804921.
DOI: 10.1055/s-0045-1804921
