Sleep and Breathing

, Volume 17, Issue 1, pp 267–274

Subjective sleepiness and daytime functioning in bariatric patients with obstructive sleep apnea


    • Department of MedicineRhode Island Hospital and Alpert Medical School of Brown University
    • Department of Psychiatry and Human Behavior, Alpert Medical SchoolBrown University
    • Division of Pulmonary, Critical Care, and Sleep Medicine; Sleep for Science Research LaboratoryBrown University/Rhode Island Hospital
  • Henry J. Orff
    • Department of PsychiatryUniversity of California, San Diego
  • Christine Tosi
    • Department of MedicineRhode Island Hospital and Alpert Medical School of Brown University
  • David Harrington
    • Department of SurgeryRhode Island Hospital and Alpert Medical School of Brown University
  • G. Dean Roye
    • Department of SurgeryRhode Island Hospital and Alpert Medical School of Brown University
  • Richard P. Millman
    • Department of MedicineRhode Island Hospital and Alpert Medical School of Brown University
Original Article

DOI: 10.1007/s11325-012-0685-3

Cite this article as:
Sharkey, K.M., Orff, H.J., Tosi, C. et al. Sleep Breath (2013) 17: 267. doi:10.1007/s11325-012-0685-3



The purpose of this study was to evaluate associations between obstructive sleep apnea (OSA) severity and self-reported sleepiness and daytime functioning in patients considering bariatric surgery for treatment of obesity.


Using a retrospective cohort design, we identified 342 patients who had sleep evaluations prior to bariatric surgery. Our final sample included 269 patients (78.6 % of the original cohort, 239 females; mean age = 42.0 ± 9.5 years; body mass index = 50.2 ± 7.7 kg/m2) who had overnight polysomnography and completed the Epworth Sleepiness Scale (ESS) and the Functional Outcomes of Sleep Questionnaire (FOSQ). Patients' OSA was classified as none/mild (apnea–hypopnea index (AHI) < 15, n = 112), moderate (15 ≤ AHI < 30, n = 77), or severe (AHI ≥ 30, n = 80). We calculated the proportion of unique variance (PUV) for the five FOSQ subscales. ANOVA was used to determine if ESS and FOSQ were associated with OSA severity. Unpaired t tests compared ESS and FOSQ scores in our sample with published data.


The average AHI was 29.5 ± 31.5 events per hour (range = 0–175.8). The mean ESS score was 6.3 ± 4.8, and the mean global FOSQ score was 100.3 ± 18.2. PUVs for FOSQ subscales showed moderate-to-high unique contributions to FOSQ variance. ESS and global FOSQ score did not differ by AHI group. Only the FOSQ vigilance subscale differed by OSA severity with the severe group reporting more impairment than the moderate and none/mild groups. Our sample reported less sleepiness and daytime impairment than previously reported means in patients and controls.


Subjective sleepiness and functional impairment were not associated significantly with OSA severity in our sample of patients considering surgery for obesity. Further research is needed to understand individual differences in sleepiness in patients with OSA. If bariatric patients underreport symptoms, self-report measures are not an adequate substitute for objective assessment and clinical judgment when evaluating bariatric patients for OSA. Patients with severe obesity need evaluation for OSA even in the absence of subjective complaints.


SleepinessObstructive sleep apneaObesityBariatric surgeryEpworth Sleepiness ScaleFunctional Outcomes of Sleep QuestionnaireSex differences


Obesity is a well-known risk factor for obstructive sleep apnea syndrome (OSA), and patients with obesity have a high prevalence of OSA [1]. For instance, OSA has been observed in 64–100 % of samples of obese patients planning bariatric surgery as a treatment for weight loss [26]. Patients with OSA are at higher risk for numerous negative health outcomes including congestive heart failure [7], coronary artery disease [8], hypertension [9], and stroke [10], as well as post-operative respiratory complications [1113], and increased perioperative morbidity and mortality following bariatric surgery [14].

Despite the risk of perioperative complications associated with untreated OSA, not all guidelines for perioperative management of patients with obesity recommend universal evaluation for sleep-disordered breathing with polysomnography, instead advocating the use of risk-stratification algorithms before proceeding with a sleep study, e.g., [15, 16]. Although several OSA screening questionnaires and prediction models have been developed, e.g., [1, 1720], there is evidence that screening based on demographics and self-reported symptoms can miss clinically significant OSA in patients planning bariatric surgery [21]. Thus, the optimal way to assess for the presence and subjective impact of OSA in bariatric patients is an issue for further study.

OSA has significant impact on mood, psychosocial functioning, and quality of life [2224]. Although patients with obesity also are known to have depressive symptoms [25], low health-related quality of life [26], and increased complaints of excessive daytime sleepiness [27], the role that excessive sleepiness may play in daily functioning of obese patients is not well understood. The Epworth Sleepiness Scale (ESS) is a validated eight-item questionnaire that measures sleep propensity by asking patients to rate their likelihood of falling asleep in various day-to-day scenarios [28]. The Functional Outcomes of Sleep Questionnaire (FOSQ) is a 30-item self-report measure that assesses the impact excessive sleepiness has on activities of daily living [29]. The FOSQ includes five subscales—intimacy, social outcomes, vigilance, activity, and productivity—that were derived from a model of functional assessment in disability [30]. The purpose of this study was (1) to evaluate sleepiness using the ESS and daytime functioning using the FOSQ in a sample of patients planning bariatric surgery for treatment of obesity and (2) to assess the association between OSA severity and self-reported tendency to fall asleep and functional outcomes of sleepiness in patients with comorbid obesity and OSA. We hypothesized that more severe OSA would be associated with higher ESS scores and more functional impairment on the global FOSQ score and on the five FOSQ subscales.


Patient selection

We identified 342 patients who were evaluated for bariatric surgery between January 2003 and December 2005 using retrospective chart review. All patients had overnight polysomnography (PSG) to be evaluated for OSA as part of mandatory preoperative screening. The patients were not pre-selected for PSG based on symptoms. Investigators reviewed the Sleep Disorders Center and office charts to collect demographic data, results of polysomnography, and ESS and FOSQ scores. From the 342 charts reviewed, we identified 269 patients with overnight PSG and complete ESS/FOSQ data. We excluded 73 patients who had PSG but did not complete the ESS/FOSQ, including 60 women and 13 men. Excluded patients did not differ from included participants in sex distribution, age, body mass index (BMI), or apnea–hypopnea index (AHI). This study was approved by the Rhode Island Hospital Institutional Review Board.

Epworth Sleepiness Scale and Functional Outcomes of Sleep Questionnaire

The ESS was our measure of subjective sleepiness. This widely used measure of the tendency to fall asleep during the day has high internal reliability (Cronbach's alpha = 0.88) and test–retest reliability [31], and has been shown to distinguish patients with OSA from those with primary snoring [28]. Patients rated their tendency to fall asleep on eight items with ratings from 0 to 3, where 0 indicated no chance of dozing and 3 indicated a high chance of dozing during the activity. Thus, scores could range from 0 to 24 with higher scores indicating more sleepiness. An ESS score ≥10 is indicative of clinically significant sleepiness.

We used the FOSQ as our measure of functional impairment due to daytime sleepiness. The FOSQ has a reported internal reliability of alpha = 0.95 and a test–retest reliability of 0.90, and has been shown to successfully distinguish patients with clinical sleep problems from healthy controls [29]. Patients rated their impairment in performing various activities due to sleepiness on a scale of 0 to 4, where 0 indicated that the person did not do this activity, 1 indicated extreme difficulty with the activity, and 4 indicated no difficulty. Thus, the possible range of the global FOSQ score was 0 to 120, with higher scores indicating higher functioning/less impairment [29]. We also calculated five FOSQ subscale scores for intimacy, social outcomes, vigilance, activity, and productivity as described by Gooneratne and colleagues [32]. Subscale scores consisted of the average rating for all items within each subscale; thus, the possible range of subscale scores was 1 to 4, with higher scores indicating higher functioning/less impairment. The ESS and FOSQ were included among the packet of questionnaires that all patients completed after they were scheduled for their sleep evaluation and were not used to determine which bariatric patients would undergo PSG.


Standard overnight polysomnography was performed at the sleep center using the Viasys sleep acquisition system (Yorba Linda, CA). Sleep was recorded using central and occipital electroencephalographic leads, bilateral electro-oculograms, and a submental electromyogram. Respiration was monitored using continuous pulse oximetry, a snoring microphone, a nasal pressure transducer, oral and nasal thermistors, and chest and abdominal piezo electrodes. Heart rate was continually monitored using a modified V2 lead, and bilateral anterior tibialis EMG leads were used to detect periodic limb movements.

All records were visually scored in 30-s epochs using Rechtshaffen and Kales criteria [33], and the Viasys software was used to tabulate sleep stages and sleep-disordered breathing indices. The same technologists did all scoring and had periodic concordance checks performed with physicians board certified in sleep medicine. Respiratory events were scored according to “Chicago” consensus criteria [34]. Apneas were defined as an absence of airflow lasting 10 s or longer. Hypopneas were scored if there was a 50 % decrease in the nasal pressure transducer signal compared to baseline that lasted at least 10 s and occurred with either a >3 % oxygen desaturation or an arousal. Events were classified as central in the absence of any respiratory effort, mixed if there was initially no respiratory effort followed by progressive evidence of ineffective respiration, and obstructive if there was persistent respiratory effort despite an absence of airflow. EEG arousals were determined using established criteria [35]. OSA severity was classified using the AHI (defined as the total number of apneas and hypopneas divided by the total number of hours of sleep). None/mild OSA was defined as AHI <15, moderate OSA was defined as ≥15 AHI <30, and severe OSA was defined as AHI ≥30 episodes per hour of sleep [34].

Statistical analyses

Analyses were performed using SPSS 19.0 (IBM, Armonk, NY) and GraphPad (GraphPad Inc., La Jolla, CA). Due to the unequal numbers of men and women in the sample, we compared sex differences in demographic variables, sleep apnea severity, medical comorbidities, ESS, global FOSQ scores, and FOSQ subscales using Student's t tests and chi-square. Internal reliability of the FOSQ in our sample was determined with Cronbach's alpha. We calculated the proportion of unique variance (PUV) for the five FOSQ subscales by subtracting the correlation of each scale with the other four scales from the scale's alpha reliability estimate. This approach of subtracting the shared variance between subscales (R2) from the total reliable variance (alpha) allows for an estimate of the unique variance provided by the individual subscales [36]. Thus, higher PUV coefficients indicate that the subscale is providing a higher proportion of unique information that is not contained in the other subscales.

ANOVA was used to determine if ESS, global FOSQ score, and/or FOSQ subscales were associated with OSA severity. We used unpaired t tests to compare ESS scores in our sample with data previously published by Johns [28, 31], including medical students (mean ESS = 7.3 ± 3.9, n = 104) and patients with snoring (RDI < 5; n = 108 events per hour; mean ESS, 8.0 ± 3.5), mild OSA (RDI = 5–24.9 events per hour; n = 105; mean, ESS = 11 ± 4.2), moderate OSA (RDI = 25–49.9 events per hour, n = 41, mean ESS = 13 ± 4.7), and severe OSA (RDI ≥ 50 events per hour, n = 19, mean ESS = 16.2 ± 3.3). Similarly, we compared global FOSQ scores in our sample with previously published scores from the original FOSQ validation study by Weaver and colleagues [29], which included a group of sleep clinic patients (mean global FOSQ = 68.1 + 21.2, n = 133) and a group of healthy controls (mean global FOSQ = 89.6 + 8.6, n = 20).



Detailed information about participants is shown in Table 1. Participants included 239 women and 30 men. Average age was 42.0 ± 9.5 years, and mean BMI was 50.2 ± 7.7 kg/m2. Most participants had at least moderate OSA. The average AHI was 29.5 ± 31.5 events per hour of sleep (range = 0–175.8 events per hour). One hundred and twelve patients were classified as none/mild (AHI < 15), 77 were considered moderate (15 ≤ AHI < 30), and 80 were classified as severe (AHI ≥ 30). Men had significantly higher average BMIs and higher AHIs and were more likely than women to have severe OSA and diabetes mellitus (see Table 1).
Table 1

Description of the sample


Total sample



t (df)

N = 269

n = 239

n = 30

p value

Age (years, mean ± SD)

42.0 ± 9.5

41.8 ± 9.5

43.3 ± 9.6

0.79 (267)


BMI (kg/m2, mean ± SD)

50.2 ± 7.7

49.7 ± 7.3

53.9 ± 9.6

2.82 (266)

p = 0.005

AHI (events/h, mean ± SD)

29.5 ± 31.5

26.5 ± 30

53.2 ± 33.6

4.53 (267)

p < 0.001




p value

Percent with diabetes mellitus

24.5 %

22.2 %

43.3 %

6.44 (1)

p = 0.011

Percent with hypertension

39.0 %

38.1 %

46.7 %

0.83 (1)


Percent in each OSA category


 None/mild (AHI < 15)

41.6 %

45.6 %

10 %

23.93 (2)

 Moderate (15 ≥ AHI > 30)

28.6 %

29.3 %

23.3 %

p < 0.001

 Severe (AHI ≥ 30)

29.7 %

25.1 %

66.7 %


t tests and chi-square show results of tests of sex differences for the sample

Epworth Sleepiness Scale

Average ESS score was 6.3 ± 4.8 (range = 0–23). Figure 1 (left column) shows the distribution of ESS scores for the three OSA severity groups. ESS scores did not differ based on AHI (see Table 2).
Fig. 1

Distribution of ESS and FOSQ scores. Plots show ESS scores (in two-unit bins) and FOSQ scores (in ten-unit bins) for none/mild, moderate, and severe OSA. There was no relationship between severity of sleep-disordered breathing and ESS (left column) and FOSQ scores (right column)

Table 2

ESS and FOSQ scores


Total sample




Fdf, p value

N = 269

n = 112

n = 77

n = 80

ESS score

6.3 ± 4.8

6.1 ± 4.3

6.0 ± 4.9

6.8 ± 5.4

F2,266 = 0.77, p = ns

Total FOSQ score

100.3 ± 18.2

101.8 + 16.6

98.2 + 20.6

100.3 + 18.0

F2,2660.89, p = ns


3.73 ± 0.46

3.78 + 0.39

3.69 + 0.48

3.71 + 0.53

F2,2661.03, p = ns


3.60 ± 0.55

3.66 + 0.45

3.65 + 0.50

3.46 + 0.68

F2,2653.66, p = 0.027

 Social outcomes

3.81 ± 0.48

3.86 + .0.39

3.76 + 0.44

3.78 + 0.61

F2,2210.27, p = ns


3.65 ± 0.64

3.61 + 0.59

3.61 + 0.79

3.73 + 0.54

F2,2660.37, p = ns


3.36 ± 0.60

3.40 + 0.54

3.32 + 0.68

3.33 + 0.60

F2,2640.20, p = ns

Average±SD ESS, global FOSQ, and FOSQ subscale scores for the entire sample and for each OSA severity group. Lower FOSQ scores indicate more impairment and poorer functioning. Patients in the severe OSA group reported significantly lower vigilance functioning on the FOSQ than those in the none/mild and moderate groups. There were no significant differences among the OSA severity groups in ESS, global FOSQ score, or any of the other subscores

Because the majority of patients in our sample were women and we were concerned this finding may be driven by sex differences, we repeated the ANOVA comparing mean ESS in female participants only, and once again we observed no relationship between OSA severity and ESS score (F2,236 = 0.064, p = 0.94).

The average ESS score for patients with moderate or severe OSA (AHI ≥ 15, n = 157) was 6.4 ± 5.2. Compared to published means [28, 31], our participants with moderate-to-severe OSA had ESS scores that did not differ significantly from the ESS scores of healthy medical students (t = 1.51(df = 259), p = 0.13). Furthermore, our moderate and severe OSA patients had ESS scores that were significantly lower than snorers (t = 2.79 (df = 263), p = 0.006) and patients with mild (t = 7.56 (df = 260), p < 0.0001), moderate (t = 7.38 (df = 196), p < 0.0001), and severe OSA (t = 8.01 (df = 274), p < 0.0001) (See Fig. 2).
Fig. 2

ESS and FOSQ scores in bariatric patients compared to published norms. The top panel shows the mean ESS score of our bariatric patients with AHI ≥15 compared to the published ESS scores of medical students [31] and sleep clinic patients with snoring and mild, moderate, and severe OSA [28]. The ESS scores of bariatric patients with moderate-to-severe OSA did not differ from the ESS scores of healthy medical students and were significantly lower than all sleep clinic patients including those with primary snoring and mild OSA. The bottom panel shows the global FOSQ scores of our bariatric patients compared to published norms from healthy controls and sleep clinic patients [29]. Our patients reported significantly higher functioning than both comparison groups. * = p < 0.05; *** = p < 0.001

Functional Outcomes of Sleep Questionnaire

In our sample, the internal reliability of the FOSQ was high (Cronbach's alpha = 0.96). The PUVs for FOSQ subscales were: intimacy = 0.53, social = 0.38, vigilance = 0.41, activity = 0.20, and productivity = 0.19, indicating that each of the subscales contributed unique variance to the total variance of the global FOSQ score.

The average global FOSQ score was 100.3 ± 18.2. Table 2 shows FOSQ scores for the total sample and for each OSA severity group. Global FOSQ score did not differ based on AHI (see Fig. 1, right column). Only the vigilance subscale differed between OSA severity groups, with the severe group demonstrating lower vigilance scores than the moderate group and the none/mild group. There were no sex differences in global FOSQ score or in any of the subscales. Because our sample was comprised mostly of women, we examined for sex differences driving this finding by repeating the ANOVA comparing mean FOSQ global and subscale scores in female participants only, and once again, we observed a relationship between OSA severity only in the FOSQ vigilance subscale (F2,237 = 3.179, p = 0.043).

One-sample t tests comparing global FOSQ scores in the present sample with average global FOSQ scores published in the original FOSQ validation paper by Weaver and colleagues [29] showed that participants in our study reported significantly fewer symptoms than either Weaver et al.'s sleep clinic patients (mean = 68.1 + 21.2, t = 29.0 (df = 268), p < 0.001) or healthy controls (mean = 89.6 + 8.6, t = 9.7 (df = 268), p < 0.001) (see Fig. 2).


Our analysis of the ESS and FOSQ in patients considering bariatric surgery for the treatment of obesity showed no relationship between OSA severity and subjective reports of daytime sleepiness or functional impairment. Our analyses showed that all five FOSQ subscales contributed independent information to the variance of FOSQ scores in this sample. Nevertheless, our hypothesis that sleepiness as measured with the ESS and daytime functioning as measured by the global FOSQ and its subscales would be associated with severity of OSA was not supported. Only the FOSQ vigilance subscale was related to severity of obstructive sleep apnea in these patients with a high prevalence of sleep-disordered breathing. Furthermore, the vigilance subscale only distinguished those patients with the most severe OSA as indicated by AHI ≥30 events per hour of sleep.

These data suggest that relying on subjective report for determining which bariatric surgery candidates require assessment for OSA does not identify all patients with significant sleep-disordered breathing. Previous studies have indicated that measures such as BMI, Epworth Sleepiness Scale, neck circumference, and waist-to-hip ratio may be helpful for predicting OSA in asymptomatic patients with obesity [3, 19, 37, 38]. On the other hand, even when statistically significant predictors are identified, they do not always perform with the precision needed for clinical utility [39]. Thus, the present data raise the question of whether performance tasks that measure vigilance may be a useful adjunct for assessment of OSA and its effects in patients with obesity.

It is striking that nearly 60 % of our sample had moderate or severe OSA that had not been diagnosed prior to the patients' evaluation for bariatric surgery. Previous studies have shown a high prevalence of OSA in patients with obesity despite absence of patient-reported symptoms, e.g., [40]. In light of the negative consequences of OSA both for general health and post-operative outcomes, bariatric providers should be aware of the high prevalence of clinically significant OSA among patients who do not report significant impairment in daily functioning from sleepiness. Clinicians should also be aware that patients who do not report or experience sleepiness or functional impairment may also have poorer treatment compliance [41].

Given the significant OSA observed in the majority of the sample, we were surprised that our patients reported less impairment in daily functioning from sleepiness than previous samples of sleep clinic patients and healthy controls. Patients planning bariatric surgery for treatment of obesity may underreport symptoms for various reasons. For instance, patients may believe that surgery will be denied if they endorse symptoms. Another possible explanation for the low reports of functional impairment is inability to recognize sleepiness and/or its effects. Cognitive deficits including such domains as executive functioning have been documented in patients with obesity [42] and specifically in those planning bariatric surgery [43]. In addition, failure to appreciate one's own sleepiness despite objective evidence of sleepiness has been shown in clinical populations [44], including patients with OSA [45]. Thus, patients simply may be unaware of their sleepiness or its impact on their functioning. Furthermore, patients with obesity may experience other weight-related symptoms (e.g., nocturia, sexual dysfunction, gastroesophageal reflux, depression, back and joint pain, exertional dyspnea) that are more noticeable or bothersome than sleepiness [23, 46]. Factors such as habitual sleep time [47], subjective sleep quality [48], socioeconomic status [49], and comorbid medical and psychiatric conditions [50] also may have mediated subjective reports of sleepiness and daytime impairment in our participants. Finally, there may be a subset of patients with OSA and obesity who are not sleepy or who do not attribute their functional impairment to sleep difficulties. Inter-individual variability in symptoms of OSA is poorly understood, and gaining a better understanding of individual differences in the association between sleep-disordered breathing and daytime functioning is an area deserving further investigation. Ultimately, it is likely that several different OSA phenotypes will be identified and described, which could lead to better understanding of OSA pathophysiology and to tailored treatments.

The major strengths of this study are our relatively large sample size and the use of the validated Epworth Sleepiness Scale and Functional Outcomes of Sleep Questionnaire to measure daytime consequences of OSA. One limitation of the study is the relative paucity of male patients compared to female patients in our sample. However, this gender imbalance reflects the overall population of patients seeking bariatric surgery, and the observed sex differences in our sample (namely higher BMI, higher AHI, and increased diabetes and severe OSA in male patients) did not affect the main findings of this analysis. Another methodologic concern is our lack of a control group—although this investigation focused on a series of bariatric surgery-seeking patients studied at our center, we sought to contextualize the FOSQ results in our sample by comparing these data to published means from other samples. Future work could investigate patients with obesity who are not considering bariatric surgery or evaluate sleepiness before and after bariatric surgery to observe possible changes in sleepiness after weight loss.

In conclusion, we found a high prevalence of moderate-to-severe OSA in our sample of 239 female and 30 male patients planning bariatric surgery. We hypothesized that patients with high levels of OSA would experience significant daytime complaints. The observation that this group showed little evidence of impairment on the ESS and the FOSQ, coupled with the fact that they scored lower on the ESS and FOSQ than previously studied clinical samples, indicates that these patients may underreport symptoms. Such underreporting (for whatever reason) should raise concerns for teams providing bariatric care because it suggests that self-report measures of impairment may be poor indicators of functioning and may not be an adequate substitute for objective assessment (e.g., polysomnography) and clinical judgment. Given the risk OSA confers on patients for complications in their planned surgery and recovery, the complex association of obesity, OSA, and daytime functioning warrants further systematic evaluation and study in bariatric populations.


The authors thank the polysomnographic technologists and clerical staff at the Sleep Disorders Center for their excellent work with our patients.


The authors declare that they have no conflict of interest.

Copyright information

© Springer-Verlag 2012