Analysis of Sleep Parameters in Patients with Obstructive Sleep Apnea Studied in a Hospital vs. a Hotel-Based Sleep Center

Abstract

Background:

Polysomnography is associated with changes in sleep architecture called the first-night effect. This effect is believed to result from sleeping in an unusual environment and the technical equipment used to study sleep. Sleep experts hope to decrease this variable by providing a more familiar, comfortable atmosphere for sleep testing through hotel-based sleep centers. In this study, we compared the sleep parameters of patients studied in our hotel-based and hospital-based sleep laboratories.

Methods:

We retrospectively reviewed polysomnograms completed in our hotel-based and hospital-based sleep laboratories from August 2003 to July 2005. All patients were undergoing evaluation for obstructive sleep apnea. Hospital-based patients were matched for age and apnea-hypopnea index with hotel-based patients. We compared the sleep architecture changes associated with the first-night effect in the two groups. The associated conditions and symptoms listed on the polysomnography referral forms are also compared.

Results:

No significant differences were detected between the two groups in sleep onset latency, sleep efficiency, REM sleep latency, total amount of slow wave sleep (NREM stages 3 and 4), arousal index, and total stage 1 sleep.

Conclusions:

This pilot study failed to show a difference in sleep parameters associated with the first-night effect in patients undergoing sleep studies in our hotel and hospital-based sleep laboratories. Future studies need to compare the first-night effect in different sleep disorders, preferably in multi-night recordings.

Citation:

Hutchison KN; Song Y; Wang L; Malow BA. Analysis of sleep parameters in patients with obstructive sleep apnea studied in a hospital vs. A hotel-based sleep center. J Clin Sleep Med 2008;4(2):119–122.

One of the limitations of polysomnography is the “first-night effect,” which is believed to result from sleeping in the unfamiliar environment of a sleep laboratory. This phenomenon is associated with alterations in sleep architecture on the first night of sleep investigation, in comparison to subsequent nights. Agnew et al. were the first to describe this intuitive observation in1966.¹ They reported that the first night in the sleep laboratory contained more wake time, less REM sleep, and increased REM and stage 4 latencies. These effects rapidly tapered by the second night of sleep. Subsequently, the first-night effect has been reported in up to 80% of healthy subjects.²

Because the first-night of laboratory polysomnography may never be able to accurately capture one's true sleep, it has become customary in sleep research to exclude data from the first night of testing. In fact, to save paper and tedious scoring time, early sleep studies often did not even record data from the first night.¹

Clinical sleep evaluations, however, are typically restricted to one night of testing due to cost and space limitations. Sleep experts hope to minimize the first-night effect associated with clinical polysomnography by controlling environmental confounds felt to contribute to the effect, namely comfort and familiarity. One way to accomplish this is through hotel-based sleep centers.

In this study, we compared the sleep parameters of patients studied in our hotel-based and hospital-based sleep laboratories. We predicted that sleep architecture changes consistent with the first-night effect would be greater in the hospital-based sleep center.

METHODS

This was a retrospective review of polysomnograms from two groups of patients undergoing sleep studies at our institution. One group had their polysomnograms at our 10-bed hotel-based laboratory in the Vanderbilt Marriott Hotel, which opened in August 2003. The second group had their polysomnograms in our two-bed hospital-based laboratory located in the basement of The Vanderbilt Clinic building (connected to the hospital). There were no major differences in the beds or equipment used in the two locations. When the new hotel-based sleep center opened, a portion of the hospital-based laboratory was maintained for patients with increased medical needs and to handle overflow from the hotel-based center. These two sleep laboratories ran simultaneously for two years before the hospital-based laboratory was closed

After receiving IRB approval, we reviewed all adult sleep studies performed in our hospital-based laboratory from August 2003 to July 2005. We excluded continuous positive airway (CPAP) titration and split-night studies, studies performed in patients assigned to the hospital-based laboratory because of medical need, and patients being evaluated for any sleep disorder other than sleep apnea. Hospital-based patients were matched with hotel-based patients for age and apnea-hypopnea index (AHI) using six months of hotel-based data (Jan.-Jun. 2004). Polysomnograms included central and occipital electroencephalogram (EEG), electrooculogram, submentalis electromyography (EMG), airflow, nasal pressure, electrocardiogram, thoracoabdominal motion, anterior tibialis EMG, and pulse oximetry. The tracings were scored in 30-sec epochs by the same scorer using standard criteria.⁷ We defined an obstructive apnea as a greater than 90% decrease from baseline in the amplitude of the thermistor channel during sleep for ≥10 seconds with continued thorocoabdominal movement. A hypopnea was defined as a clear amplitude reduction of 50%–90% in either the thermistor or the nasal pressure channel during sleep that was associated with either an oxygen desaturation of ≥3%, or an EEG arousal.⁸ Apnea/hypopnea index (AHI) was defined as the total number of episodes of apnea and hypopnea per hour of sleep.

We compared the following sleep parameters between the two groups: sleep efficiency, sleep onset latency, REM sleep latency, total amount of slow wave sleep (NREM stages 3 and 4), total stage 1 sleep, and arousal index.

In addition, we reviewed all available sleep study referral forms and compared the listed medical histories and associated conditions.

All statistical analyses were performed using the software R Version 2.3.1.⁶ and SAS Version 9.13 (SAS Institute Inc, Cary, NC). A 2-sided P value <0.05 was considered statistically significant. Because of the skewness in data, sleep parameters and clinical variables for patients at the two locations were compared using the nonparametric Wilcoxon rank-sum test.

Based on the variations of sleep latency between subjects in these data, we next estimated the sample sizes needed to detect a clinical meaningful difference of 15 min in sleep latency for the two groups. The results showed a sample size of 45 patients in each group will provide 80% power to detect a difference in means of 15 min in sleep latency for hospital-based and hotel-based patients, using a two group t-test with a 0.05 two-sided significance level, assuming the standard deviation of sleep latency for each group does not exceed 25 minutes. The power of the current study with 44 and 49 patients in the two groups gave us about 79% power to detect the 15-min difference in sleep latency.

RESULTS

A total of 410 studies were reviewed with 44 hospital-based studies meeting our criteria. These were matched with 49 hotel-based patients. Because of the skewness in this set of data, we present results of patient characteristics in terms of 25^th percentile, median, and 75^th percentile in Table 1. The hotel-based patients and hospital-based patients had similar ages and AHIs.

Table 1.

Percentiles of Demographic and Clinical Characteristics for Hotel-Based and Hospital-Based Patients

Variable	Hotel (N=49)			Hospital (N=44)			p - value**
	25th	Median	75th	25th	Median	75th
Age	37.0	45.0	56.0	40.0	46.5	52.5	0.99
AHI	5.7	9.7	22.0	5.2	9.2	14.9	0.50
REM Latency	74.0	111.0	232.0	66.8	102.0	266.0	0.86
Sleep Latency	7.0	16.0	23.0	7.0	12.0	29.5	0.97
Sleep Efficiency. %	68.4	82.9	91.4	76.9	80.7	88.3	0.76
SWS	0.0	0.1	7.1	0.0	0.0	4.2	0.83
Stage 1*	8.9	10.3	14.0	8.2	11.4	16.1	0.99
Arousal Index	6.8	9.8	18.4	5.8	11.0	14.7	0.79

^*Sample size for stage 1 is 35 for Hotel and 42 for Hospital

^**Based on Wilcoxon rank-sum test

The hotel-based patients and hospital-based patients did not differ in any of the sleep parameters. In particular, the two groups had similar sleep efficiency, sleep onset latency,

REM sleep latency, slow wave sleep, arousal index and stage 1 sleep as listed in Table 1.

Sleep study referral forms were available for 48 of the hotel-based subjects and 42 of the hospital-based subjects. Descriptive variables are listed in Tables 2 and 3. No significant differences were noted in the covariate data.

Table 2.

Comorbid Medical Diagnoses Listed on PSG Referral Form

	Hotel-based (N=49)	Hospital-based (N=44)	p-value
Lung Disease (COPD – Asthma)	7(14.29%)	11(25.00%)	0.19
Hypertension	12(24.49%)	13(29.55%)	0.58
Heart disease (CAD, CHF, Arrhythmia)	10(20.41%)	4(8.33%)	0.13
Diabetes	7(14.29%)	5(10.42%)	0.67
Chronic allergies, sinusitis	7(14.29%)	3(6.25%)	0.14
GERD	5(10.20%)	3(6.25%)	0.25
Neurological disorder (migraine, MD)	7(14.29%)	3(6.25%)	0.14
Psychiatric disorder (anxiety, depression)	11(22.45%)	8(16.67%)	0.61
None	11(22.45%)	9(20.45%)	0.97

COPD = chronic obstructive pulmonary disease, CHF = congestive heart failure, CAD = coronary artery disease, GERD = gastrointestinal reflux disease, MD = myotonic dystrophy.

Table 3.

Associated Symptoms Listed on PSG Referral Form

	Hotel-based (N=49)	Hospital-based (N=44)	p-value
Excessive daytime sleepiness	37(75.51%)	25(56.82%)	0.06
Insomnia	3(6.12%)	2(4.17%)	0.34
Impaired cognition	2(4.02%)	1(2.08%)	0.40
None	9(18.37%)	16(36.36%)	0.051

DISCUSSION

To our knowledge, this is the first study to compare sleep parameters in hotel vs. hospital-based sleep centers. Our sample provided a unique opportunity to compare patients studied in hospital and hotel-based labs that were drawn from the same population, and had comparable PSG equipment and protocols, technologists, and scoring algorithms applied.

This pilot study failed to show a significant difference in sleep parameters in our hotel and hospital-based sleep laboratories. In fact, the results of the two groups were strikingly similar. These results counter our hypothesis that creating a more familiar environment, such as that offered by a hotel room, may reduce or eliminate the sleep architecture changes associated with the first-night effect. Despite minimal differences between the two groups, sleep architecture abnormalities were detected in both groups as compared to standard norms,³ with reduced sleep efficiency and prolonged REM and sleep latencies noted.

The first night effect phenomenon originated in early sleep research when the sleep laboratory was commonly housed in a hospital basement or research corridor. The environment was often stark and unfriendly and quite unfamiliar to the patients. Sleep research, and sleep laboratories, however, have made significant improvements over time. One potential reason why no difference was detected may be that the hotel and hospital environments were too similar. While our hospital-based laboratory was located in the basement of our clinic building adjacent to the hospital, the rooms were furnished with hotel-like amenities such as carpet, a large TV, and hotel-style linens.

If sleep parameters associated with the first-night effect are the same in both a hotel and a hospital room, perhaps the effect has more to do with sleeping away from home or the testing procedure itself. EEG wires, nasal thermistors and the feeling of immobility can all contribute to disrupted sleep. These are not unique to the location. In-home polysomnography, however, has shown less dramatic changes between first and subsequent nights of testing.³ This suggests that the unfamiliarity of being away from home may indeed contribute to more dramatic sleep architecture changes.

The American Academy of Sleep Medicine, the primary governing body of sleep disorders centers, lists in the Standards for Accreditation of Sleep Disorders Centers that “patient testing rooms must afford comfort, privacy, safety, and accessibility…”⁴ Rising demands for these testing rooms combined with limited hospital space (particularly in academic centers) has forced sleep centers to expand outside of their traditional testing locations. More and more sleep laboratories are finding that hotels fulfill the necessary requirements for polysomnography while providing consistent occupancy in a competitive hotel market. But does a more familiar, user-friendly atmosphere accomplish the goal of obtaining more accurate sleep data? This study suggests not.

Hotel-based sleep laboratories, however, do offer several advantages. They are typically more convenient with adequate parking and friendly, available staff. The hotel environment is generally more familiar and less intimidating to patients, often providing additional amenities such as restaurants, workout facilities, and swimming pools. Hotel space is significantly less expensive than hospital space, and is more widespread. These advantages have led to an explosion of hotel-based sleep labs.

One limitation of this study is the small sample size. Because the hospital-based sleep laboratory was ultimately closed, no additional data are available. Also, the large between-subjects variability for some of the sleep parameters may render the statistical tests less powerful. We have based power calculations on sleep latency, as this is the sleep parameter we are most interested in; however, for other parameters with larger between-patients variability, a larger sample size may be needed to have enough power to detect clinical meaningful differences between hotel-based and hospital-based patients. In addition, comparing repeated measures from multi-night recordings would provide a more accurate evaluation of the first-night effect and a more sensitive way of detecting potential differences between the two groups.

Another limitation is that the first-night effect was originally described in healthy, young subjects without sleep related complaints. There is a need to compare the first-night effect of patients with sleep disorders to age-matched, normal controls. It may be that the symptoms which led to the initial sleep evaluation, e.g., excessive daytime sleepiness, may blunt the hyperaroused state detected in normal sleepers on the first night of testing. The addition of a standardized, subjective measure of sleepiness and/or pre-testing actigraphy would provide a measure of sleep pressure, which intuitively would affect the first-night effect. The variability in sleep pressure was minimized in this study by selecting only patients being evaluated for obstructive sleep apnea in both settings.

In future studies, larger groups of subjects with a variety of sleep disorders need to be studied, preferably in multi-night recordings. Overall patient satisfaction and cost effectiveness should also be evaluated. While our results document no differences in sleep architecture between our two labs, it is unclear whether our results are generalizable to other hospital and hotel-based labs. Further studies will be necessary to determine if this is the case.