Patients with COPD run a risk of developing. Underestimation of Nocturnal Hypoxemia Due to Monitoring Conditions in Patients With COPD*

Transcription

1 Underestimation of Nocturnal Hypoxemia Due to Monitoring Conditions in Patients With COPD* Folkert Brijker, MD; Frank J. J. van den Elshout, MD, PhD; Yvonne F. Heijdra, MD, PhD; and Hans Th. M. Folgering, MD, PhD Study objectives: COPD patients run a risk of developing nocturnal oxygen desaturation. When evaluating patients with nocturnal hypoxemia, an unfamiliar hospital environment and the monitoring equipment may cause sleep disturbances. It was hypothesized that increased sleep disruption will lead to fewer instances of desaturation during a night of monitoring. Design:The following forms of monitoring were evaluated prospectively on 3 nights for each patient: oximetry at home; polysomnography (PSG) at home; and PSG in the hospital. Setting: Department of Pulmonology, Rijnstate Hospital Arnhem, The Netherlands. Patients: Fourteen stable COPD patients (7 men; median age, 71.5 years; age range, 59 to 81 years; FEV 1, 32.5% predicted; FEV 1 range, 19 to 70% predicted) participated in the study. All subjects had significant instances of nocturnal arterial oxygen desaturation. Those patients with a sleep-related breathing disorder or cardiac failure were excluded from the study. Measurements and results: The mean nocturnal arterial oxygen saturation (SaO 2 ) level was higher during PSG monitoring at home (89.7%; range, 77 to 93%) than during oximetry monitoring (88.5%; range, 80 to 92%) [p < 0.025]. The fraction of time spent in hypoxemia (ie, SaO 2 < 90%) was lower during PSG monitoring at home (40.8%; range, 5 to 100%) than during oximetry monitoring (59.9%; range, 6 to 100%) [p < 0.01]. Desaturation time ( SaO 2 > 4%) was lower during PSG monitoring at home (22.1%; range, 3 to 63%) during PSG monitoring at home than during oximetry monitoring (50.4%; range, 4 to 91%) [p < 0.01]. A correction for actual sleep during PSG monitoring reduced the differences between PSG monitoring at home and oximetry monitoring, although a difference in the desaturation time remained (PSG monitoring at home, 31.9% [range, 2 to 75%]; oximetry monitoring, 50.4% [range, 4 to 91%]) [p 0.041]. A comparison of sleep architectures for nights when PSG was being monitored showed a higher arousal index in the hospital than at home (PSG monitoring in the hospital, 5.6 arousals per hour [range, 2 to 16 arousals per hour]; PSG monitoring at home, 2.5 arousals per hour [range, 1 to 6 arousals per hour]) [p < 0.025], but no differences in SaO 2 levels were found between PSG monitoring at home and PSG monitoring in the hospital. Conclusion: The artifacts due to sleep-monitoring equipment may cause an underestimation of the degree of nocturnal hypoxemia in COPD patients. The addition of an unfamiliar environment causes more sleep disruption, but this does not affect nocturnal SaO 2 levels further. (CHEST 2001; 119: ) Key words: COPD; nocturnal hypoxemia; oximetry; polysomnography Abbreviations: OSAS obstructive sleep apnea syndrome; PSG polysomnography; REM rapid eye movement; Sao 2 arterial oxygen saturation *From the Department of Pulmonology (Drs. Brijker and van den Elshout), Rijnstate Hospital, Arnhem, The Netherlands; and the Department of Pulmonology (Drs. Heijdra and Folgering), Dekkerswald, University of Nijmegen, The Netherlands. This study was supported by GlaxoWellcome, The Netherlands. Manuscript received April 25, 2000; revision accepted November 29, Correspondence to: Folkert Brijker, MD, Department of Pulmonary Diseases, Rijnstate Hospital, PO Box 9555, 6800 TA Arnhem, The Netherlands; zonnebloem@compuserve.com Patients with COPD run a risk of developing nocturnal hypoxemia. This can be explained by alveolar hypoventilation 1 3 and by ventilation/perfusion mismatch. 4,5 Alveolar hypoventilation occurs predominantly during rapid eye movement (REM) sleep. Ventilation/perfusion mismatch can occur during non-rem sleep episodes as well. The relevance of evaluating nocturnal hypoxemia in patients with COPD is still under debate and, as yet, no studies have shown the benefit of correcting isolated hypoxemia on long-term mortality rates. However, evaluation of the degree of nocturnal hypoxemia can be needed, as it clearly seems to be associated with complications such as arrhythmia, polycythemia, pulmonary hypertension, and peripheral edema. 6 Fur Clinical Investigations

2 thermore, the correction of nocturnal hypoxemia by supplemental oxygen therapy was found to reduce pulmonary artery pressure in a study 7 evaluating COPD patients whose daytime Pao 2 was 60 mm Hg. It is important to assess the severity of nocturnal hypoxemia in COPD patients accurately, as a value of 30% of time spent in hypoxemia is causally related to permanent pulmonary hypertension. 8 When measuring nocturnal oxygen saturation, the monitoring conditions may cause disturbances in sleep architecture. Studies on polysomnography (PSG) show a delayed sleep onset, sleep fragmentation, frequent arousals, and shortened periods of REM sleep compared to a normal sleep cycle in a comparable age group It is unclear to what extent these sleep disturbances can be attributed to monitoring equipment or to an unfamiliar hospital environment. Because of these sleep disturbances, the first monitoring night often is used for acclimatization. In patients with COPD, however, repeated measurements during consecutive and nonconsecutive nights in the sleep laboratory show a similar change in sleep architecture and nocturnal oxygen saturation. 2,13,14 If sleep is disturbed, wakefulness and arousals may prevent oxygen desaturations in patients with COPD. As a result, an overestimation of the mean nocturnal oxygenation may be found. It was hypothesized that both monitoring equipment and an unfamiliar hospital environment may have a disturbing effect on sleep, resulting in fewer instances of oxygen desaturation. To distinguish both effects, the following three conditions were evaluated: oximetry at home; PSG at home; and PSG in the hospital. First, it was expected that the addition of polysomnographic equipment in the home situation would result in fewer instances of desaturation, compared to monitoring with oximetry alone. Second, it was expected also that the unfamiliar environment during PSG monitoring in the hospital would result in fewer instances of desaturation than during PSG monitoring at home. Study Design Materials and Methods The effects of sleep monitoring conditions were evaluated in 14 patients with COPD during 3 nights in the following order: (1) oximetry at home; (2) PSG in the hospital sleep laboratory; and (3) PSG at home. An interval of at least 2 days was maintained between the different nights in order to rule out compensatory sleep after sleep deprivation. The median duration of the study was 24 days (range, 6 to 36 days). Alcohol and coffee were prohibited on the evening of the monitoring nights. Medication remained unchanged during the course of the study. Subjects The study was performed with 14 stable COPD patients, who were selected consecutively from the outpatient department. COPD was defined according to the standards of the American Thoracic Society. 15 All subjects had a smoking history of 10 pack-years (smokers, 3 patients; ex-smokers, 11 patients). Other characteristics of the study population are presented in Table 1. The stability of the disease was defined as a fluctuation in FEV 1 of 10% in the preceding 3 months and an absence of exacerbation in the preceding 8 weeks before the study began. None of the patients were oxygen-dependent. Subjects with a history of obstructive or central apneas or an overlap syndrome were excluded. Those with a history of cardiac failure also were excluded, and an ultrasound of the heart was performed to rule out current cardiac failure. Subjects were visited for oximetry at home (n 60). The hypothesized effects of the monitoring conditions could be evaluated only if nocturnal oxygen desaturation was present, which, consequently, left room for improvement. Therefore, patients continued participation in the study if their mean nocturnal arterial oxygen saturation (Sao 2 ) level was 92% (n 17). The other subjects were excluded. The second monitoring night consisted of PSG monitoring performed in the hospital. Subjects were excluded if sleep apnea/hypopnea syndrome, which was defined as an apnea/hypopnea index of 15 incidents per hour, was found. 16 Although they were not clinically suspect, three patients needed to be excluded at this stage of the study. One patient had obstructive sleep apnea syndrome (OSAS), one patient had an obstructive hypopnea syndrome, and one patient had a central apnea syndrome. The remaining subjects completed the study with a third monitoring night using PSG at home (n 14). The study protocol was approved by the hospital ethics committee. A written informed consent was obtained from all subjects. Measurements Oximetry was performed at night using the same portable pulse oximeter (model 8500M; Nonin Medical Inc; Plymouth, MN). Values were measured by a finger probe. Subjects were instructed to switch the oximeter on and off at the same time each night to match the recorded time during the 3 nights. Sao 2 values were stored in memory and processed afterward using appropriate software (NN856 03/95; PROFOX Associates Inc; Escondido, CA). Characteristics Table 1 Subject Characteristics* Values Sex, No. Male 7 Female 7 Age, yr 72 (59 81) Height, cm 164 ( ) Weight, kg 73.0 (61 100) BMI 26.8 ( ) FEV 1, % predicted 32.5 (19 70) FEV 1 /VC, % 45.5 (23 70) Daytime Paco 2,mmHg 46.5 ( ) Daytime Pao 2,mmHg 64.5 ( ) Daytime Sao 2,% 91.5 (78 95) Sleep apnea/hyponea 0.3 (0 3.2) index, No./h *Values given as median (range) unless otherwise noted. BMI body mass index; VC vital capacity. CHEST / 119 / 6/ JUNE,

3 PSG in the hospital (Voyageur; Nicolet Biomedical Inc; Madison, WI) was performed in the sleep laboratory. The EEG, the electro-oculogram, and the mental electromyogram were recorded to determine sleep states and were assessed manually by the neurology laboratory technicians according to the guidelines of Rechtschaffen and Kales. 17 Respiration was monitored by thoracoabdominal movements, using inductive plethysmography (model 1482; Pro-Tech Services Inc; Woodinville, WA) and by airflow thermistors (Infinity sensor 1450; New Life Technologies Inc; Midlothian, VA). Oximetry was performed simultaneously. PSG at home (model 9000; Oxford Instruments Medical Systems; Abingdon, Oxfordshire, UK) included EEG, electromyogram, electro-oculogram, and simultaneous oximetry. Thoracoabdominal movements and airflow were not measured at home. The severity of nocturnal hypoxemia was evaluated by the following primary variables: the mean nocturnal Sao 2 value; the fraction of time spent in hypoxemia (ie, Sao 2, 90%), and the fraction of time spent in desaturation. A desaturation was defined as a decrease in Sao 2 of 4% from the first 5 min of the recording (baseline awake). 14 The baseline asleep was defined as the first 5 min of continuous sleep after the first onset of sleep state II. Arousals were defined according to the preliminary recommendations of the American Academy of Sleep Medicine (formerly known as the American Sleep Disorders Association). 18 Because no sleep states were measured during the night of oximetry monitoring, Sao 2 variables were only presented for the entire night. The assessment of the sleep states during the nights of PSG allowed a calculation of these variables during actual sleep (REM and non-rem sleep) and during REM sleep specifically as well. If the mean Sao 2 level during the entire night differed between the oximetry and PSG values, an additional comparison was performed between the PSG values during actual sleep and the oximetry values for the entire night to determine whether this correction of the PSG results allowed a more reliable assessment of nocturnal Sao 2 values. The daytime Pao 2, the daytime Sao 2, and the baseline awake Sao 2 were correlated to the mean nocturnal Sao 2 value during different nights in order to evaluate the predictive value of these variables for assessing the degree of nocturnal hypoxemia. Statistical Analysis For statistical analysis, an appropriate software package (SPSS, version 6.1; SPSS Inc; Chicago, IL) was used. Because we aimed to evaluate the equipment effect (ie, oximetry vs PSG at home) and the environmental effect (ie, PSG at home vs PSG in the hospital), differences between 2 nights were compared selectively by paired t tests or by Wilcoxon signed rank tests if not normally distributed. A p value was considered to be statistically significant, because two selective comparisons were performed. p Values between and 0.05 are presented as true values. A Pearson correlation was used to perform bivariate correlation analysis, or a Spearman correlation was used if values were not normally distributed. All results are expressed as median (range). Results Oximetry vs PSG at Home The mean Sao 2, the fraction of time spent in hypoxemia, and the fraction of time spent in desaturation were statistically significantly different, showing lower Sao 2 values during oximetry monitoring than during PSG monitoring at home (Table 2). Thirteen patients spent 30% in hypoxemia during oximetry monitoring, compared to 9 patients during PSG monitoring at home. The fraction of time spent in desaturation decreased by more than half during PSG monitoring (22.1%; range, 3 to 63%) compared to that during oximetry monitoring (50.4%; range, 4 Table 2 Nocturnal SaO 2 Values for Different Nights* Variables OX PSGH PSGHOS OX vs PSGH p Value PSGH vs PSGHOS Entire night, % Baseline awake Sao (88 97) 92.1 (82 94) 93.1 (87 95) NS NS Mean Sao (80 92) 89.7 (77 93) 89.9 (79 94) NS Fraction of time in hypoxemia 59.9 (6 100) 40.8 (5 100) 39.3 (0 100) 0.01 NS Fraction of time in desaturation 50.4 (4 91) 22.1 (3 63) 16.6 (0 92) 0.01 NS Lowest Sao 2 73 (40 85) 74 (51 85) 78 (40 86) NS NS Actual sleep, % Baseline asleep Sao (75 93) 90.0 (79 94) NS Mean Sao (75 92) 88.5 (78 93) NS Fraction of time in hypoxemia 47.1 (6 100) 69.1 (5 100) NS Fraction of time in desaturation 31.9 (2 75) 25.5 (1 94) NS Lowest Sao 2 76 (51 85) 78 (53 86) NS REM sleep, % Mean Sao (70 91) 86.8 (77 90) NS Fraction of time in hypoxemia 71.4 (16 100) 82.8 (37 100) NS Fraction of time in desaturation 44.8 (2 99) 73.1 (37 100) Lowest Sao 2 78 (51 86) 79 (60 86) NS *Values are given as median (range). Sao 2 values were not applicable for oximetry (see text). OX oximetry; PSGH PSG monitoring at home; PSGHOS PSG monitoring in the hospital; NS not significant Clinical Investigations

4 Figure 1. The fraction of time spent in hypoxemia during oximetry and PSG monitoring at home for the individual patients. The thin lines represent an individual subject, and the thick line represents the median of the group. to 91%) [p 0.01]. The individual values of the time in hypoxemia and the time in desaturation are presented in Figures 1 and 2, respectively. The Sao 2 values were corrected for actual sleep during PSG monitoring at home and were compared to those for the entire night of oximetry monitoring (Table 3). In this recalculation, the mean Sao 2 value showed no statistically significant difference. The fraction of time spent in desaturation remained lower during PSG at home, but this did not reach statistical significance (p 0.041). PSG at Home vs PSG in the Hospital The PSG recordings allowed a detailed description of the sleep architecture and demonstrated more arousals in the hospital than at home (hospital, 5.6 arousals per hour [range, 2 to 16 arousals per hour]; home, 2.5 arousals per hour [range, 1 to 6 arousals per hour) [p 0.025; Table 4]. A more disturbed sleep architecture also was suggested from the increased total number of arousals and the lower Figure 2. The fraction of time spent in desaturation during oximetry and PSG monitoring at home for the individual patients. The thin lines represent an individual subject, and the thick line represents the median of the group. amount of time spent in REM sleep, but these did not reach complete statistical significance (p and p 0.041, respectively). No statistically significant differences in Sao 2 values were found during the entire night or during actual sleep (Table 2). However, when evaluating REM sleep, a higher fraction of time in desaturation was found when monitoring took place in the hospital (73.1%; range, 37 to 100%) vs home (44.8%; range, 2 to 99%) [p 0.025]. Correlation of Daytime and Nocturnal Oxygenation The correlations between daytime oxygenation variables and nocturnal Sao 2 are shown in Table 5. The strongest correlation was found between the baseline awake Sao 2 and the mean nocturnal Sao 2. Discussion The purpose of this study was to determine the effect of sleep monitoring conditions on the assess- Table 4 Sleep Architecture During PSG* Table 3 Nocturnal SaO 2 During the Entire Night of Oximetry Monitoring and During Actual Sleep With PSG Monitoring at Home* Variables OX PSGH p Value Mean Sao 2,% 88.5 (80 92) 89.2 (75 92) NS Fraction of time in 59.9 (6 100) 47.1 (6 100) NS hypoxemia, % Fraction of time in desaturation, % 50.4 (4 91) 31.9 (2 75) Lowest Sao 2,% 73 (40 85) 76 (51 85) NS *Values given as median (range), unless otherwise indicated. See Table 2 for expansion of abbreviations not used in text. Variable PSGH PSGHOS p Value Sleep latency, min 23.5 (12 127) 50.0 (3 101) NS Actual sleep time, h 5.5 (3 8) 6.3 (2 8) NS Arousals, total No (5 28) 21.0 (6 108) Arousal index, No./h 2.5 (1 6) 5.6 (2 16) Sleep efficiency, % 66.0 (46 96) 72.0 (43 95) NS Sleep states during sleep, % Non-REM sleep % of time I II 66.8 (60 89) 64.3 (55 89) NS III IV 16.1 (3 29) 19.9 (4 30) NS REM sleep, % of time 17.6 (5 24) 11.5 (7 27) *Values given as median (range). See Table 2 for expansion of abbreviations not used in text. CHEST / 119 / 6/ JUNE,

5 Table 5 Correlation Between Daytime and Nocturnal Oxygenation* OX PSGH PSGHOS Variables r p Value r p Value r p Value DPao 2 vs NSao NS DSao 2 vs NSao Baseline awake vs NSao *DPao 2 daytime Pao 2 ; DSao 2 daytime Sao 2 ; NSao 2 nocturnal Sao 2. See Table 2 for expansion of abbreviations not used in the text. ment of nocturnal Sao 2 values in patients with COPD. Equipment had a significant impact, as lower Sao 2 values were found during oximetry monitoring than during PSG monitoring at home. The addition of an unfamiliar environment seems to have no further impact, as no significant differences in Sao 2 values were found between PSG monitoring at home and that performed in the hospital. It was hypothesized that monitoring equipment would cause sleep disturbances and, consequently, would prevent oxygen desaturations. Polysomnographic equipment was added to oximetry in the familiar home situation and these were compared to oximetry alone. A disturbed sleep was found during PSG monitoring, which is in agreement with the hypothesis. The median sleep time was 5.5 h. This is rather low, taking into account a normal sleep time of 6 to 6.5 h at this age. 19 The median fraction of time spent in REM sleep was 17.6%, as compared to 20 to 23%, which is common at this age in normal subjects. 19 As expected, the nocturnal Sao 2 value was lower during oximetry monitoring than during PSG monitoring. The differences in the mean Sao 2 values were rather small, but the time spent in desaturation decreased by more than half from 47.5% (oximetry) to 20.9% (PSG). The Sao 2 values during PSG monitoring were corrected for actual sleep, to determine whether this resulted in a more reliable assessment of nocturnal Sao 2. The modified Sao 2 values approximated the data attained for the entire night of oximetry monitoring. Although not completely statistically significant, the fraction of time spent in desaturation remained higher during oximetry monitoring, which may suggest the presence of longer periods of REM sleep or of non-rem sleep states III and IV during home oximetry monitoring. A correlation analysis was performed between the daytime Pao 2 and Sao 2 values and the baseline awake Sao 2 and the mean nocturnal Sao 2 values in order to evaluate the predictive value of these variables for assessing the degree of nocturnal hypoxemia. The predictive values of daytime Pao 2 and Sao 2 for assessing the degree of nocturnal hypoxemia were statistically significant but rather poor, which is in agreement with earlier studies. 20,21 The correlation between the baseline awake Sao 2 and the mean nocturnal Sao 2 values was stronger for all evaluated nights. This may suggest that a short recording made with the patient in bed in the evening provides a better prediction for assessing the degree of nocturnal hypoxemia. It was also hypothesized that the addition of an unfamiliar hospital environment would influence the assessment of nocturnal Sao 2 levels. PSG monitoring in the unfamiliar hospital environment was, therefore, compared to PSG monitoring at home. Sleep was more disturbed in the hospital than at home, as indicated by the higher arousal index. A higher total number of arousals and a lower fraction of time spent in REM sleep also were found during PSG were in the hospital, but these differences did not reach complete statistical significance. As oxygen desaturations occur mainly during REM sleep, 1 3 a lower fraction of time spent in REM sleep was expected to result in fewer desaturations and higher Sao 2 values for monitoring performed in the hospital. However, the mean nocturnal Sao 2 values during the entire night or during actual sleep were not different between monitoring performed in the hospital and that performed at home. Apparently, the variation in sleep architecture was too small to cause significant differences in mean Sao 2 values. This is obvious, when it is taken into account that PSG monitoring in the hospital comprised more equipment than that performed at home. Monitoring performed in the hospital included measurements of thoracoabdominal movements and airflow as well. It appears that the addition of an unfamiliar environment had no further artificial effects on the recorded Sao 2 values, even when Sao 2 is evaluated in combination with an expanded set of equipment. A number of portable sleep study systems have been validated for use with patients who have obstructive sleep apneas or hypopneas, mainly because of benefits in time and cost. 22 Although not all devices record sleep quality, OSAS patients may sleep better at home than in the sleep laboratory. 23 In contrast to previous studies evaluating patients with sleep apneas or hypopneas, the present study 1824 Clinical Investigations

6 describes patients with COPD. The disturbed quality of sleep during PSG monitoring corresponds well with previous findings. 9 12,23 It is interesting to note that poor quality of sleep was reported as a common feature in COPD patients The authors note that the sleep disturbances may be caused by other factors, such as hypoxemic stress. The present results clearly show lower nocturnal Sao 2 values during oximetry monitoring than during PSG monitoring. This suggests a disturbing effect of monitoring equipment on the quality of sleep, especially on the amount of REM sleep. Hospital environmental factors were found to disturb sleep. 28 In agreement with this, our results show more sleep disturbances during monitoring in the hospital than at home. The study nights occurred in a fixed sequence. However, a potential order effect certainly cannot explain the differences shown in Sao 2 values between the study nights. Subjects may have experienced some general unfamiliarity with the study, causing some sleep disruption, mainly during the first night. Still, most desaturations were found during oximetry monitoring, which occurred on the first night of study in our design. A randomized order, with oximetry performed during the second or the third night, could possibly have resulted in an improved sleep comfort and even more desaturations. A first-night effect, however, was not found in a study 14 evaluating COPD patients during consecutive and nonconsecutive nights. It is unlikely that the sequence of the nights influenced the measured findings, but the findings may warrant a future study that would be conducted in a randomized design with a larger group of COPD patients. PSG monitoring is not a routine investigation in COPD patients to evaluate the degree of nocturnal hypoxemia, as no studies have shown the benefit of correcting isolated nocturnal hypoxemia on longterm outcome. Nevertheless, nocturnal hypoxemia in COPD can be associated with complications, such as arrhythmia, polycythemia, pulmonary hypertension, or peripheral edema. 6 Because supplemental oxygen therapy reduces pulmonary artery pressure in COPD patients with isolated nocturnal hypoxemia, recognizing patients at risk can be important. 7 A value of 30% of time spent in hypoxemia proved to be causally related to permanent pulmonary hypertension in COPD. 8 Thirteen of our subjects spent 30% of their time in hypoxemia during oximetry monitoring, compared to 9 patients during PSG monitoring at home, which suggests that 4 of the 13 subjects (31%) at risk would have been missed by looking at PSG monitoring. Therefore, the performance of home oximetry monitoring as the method of choice for evaluating the degree of nocturnal hypoxemia in COPD patients is of clinical importance. It can be performed easily, it is cheap, and, as this study shows, it provides reliable information. A wider use of nocturnal home oximetry monitoring may reveal many COPD patients with severe nocturnal hypoxemia who are unknown or have not been detected by PSG monitoring. Moreover, home oximetry monitoring may serve as an appropriate tool for evaluating the long-term benefit of treating nocturnal hypoxemia. One needs to consider that full PSG monitoring is still needed, as patients experiencing an overlap syndrome (ie, the simultaneous presence of COPD and OSAS) are at high risk for developing respiratory insufficiency and pulmonary hypertension. 6,29 Regarding the fact that three patients needed to be withdrawn from our study, it appears that a sleep disorder is hard to rule out in the absence of clear clinical evidence. The Sao 2 values probably will be higher during PSG monitoring. A correction for actual time asleep may allow a more reliable, but still incomplete, assessment of nocturnal Sao 2 levels. In conclusion, when monitoring nocturnal Sao 2 levels in patients with COPD, the impact of the sleeping conditions needs to be taken into account. It seems that monitoring equipment has a substantially more confounding effect on the assessment of nocturnal Sao 2 levels than does an unfamiliar environment. Sleep disturbances can prevent oxygen desaturation and can cause an underestimation of the degree of nocturnal hypoxemia in patients with COPD. ACKNOWLEDGMENT: We thank Frans L. A. Willekens, MSc, and Theo M. de Boo, MSc, for their advice on the statistical analysis. References 1 Fletcher EC, Scott D, Qian W, et al. Evolution of nocturnal oxyhemoglobin desaturation in patients with chronic obstructive pulmonary disease and a daytime Pao 2 60 mm Hg. Am Rev Respir Dis 1991; 144: Stradling JR, Lane DJ. Nocturnal hypoxemia in chronic obstructive pulmonary disease. Clin Sci 1982; 64: Becker HF, Piper AJ, Flynn WE, et al. Breathing during sleep in patients with nocturnal desaturations. Am J Respir Crit Care Med 1999; 159: Fletcher EC, Gray BA, Levin DC. Nonapneic mechanisms of arterial oxygen desaturation during rapid-eye movement sleep. J Appl Physiol 1983; 54: Hudgel DW, Martin RJ, Capehart M, et al. Contribution of hypoventilation to sleep oxygen desaturation in chronic obstructive pulmonary disease. J Appl Physiol 1983; 55: Weitzenblum E, Chaouat A, Charpenier C. Sleep-related hypoxemia in chronic obstructive pulmonary disease: causes, consequences and treatment. Respiration 1997; 64: Fletcher EC, Luckett RA, Goodnight-White S, et al. A double blind trial of nocturnal supplement oxygen for sleep desaturation in patient with chronic obstructive pulmonary disease and a daytime Pao 2 60 mm Hg. Am Rev Respir Dis 1992; 145: CHEST / 119 / 6/ JUNE,

7 8 Levi-Valensi P, Weitzenblum E, Rida Z, et al. Sleep-related oxygen desaturation and daytime pulmonary hemodynamics in COPD patients. Eur Respir J 1992; 5: Agnew HW, Webb WB, Williams RL. The first night effect: an EEG study of sleep. Psychophysiology 1966; 2: Mendels J, Hawkins DR. Sleep laboratory adaptation in normal subjects and depressed patients ( first night effect ). Electroencephalogr Clin Neurophysiol 1967; 22: Coble P, McPartland RJ, Sila WJ, et al. Is there a first night effect (a revisit). Biol Psychiatry 1974; 9: Aber WR, Block AJ, Hellard DW, et al. Consistency of respiratory measurements from night to night during the sleep of elderly men. Chest 1989; 96: Gothe B, Hanekamp LM, Cherniack NS. Reproducibility of ventilatory measurements during sleep on different nights in patients with chronic obstructive pulmonary disease. J Lab Clin Med 1987; 109: Vos PJE, Folgering HThM, van Herwaarden CLA. Sufficient indication of nocturnal oxygen saturation and breathing pattern in COPD patients, from a single night s study. Respir Med 1995; 89: American Thoracic Society. Standards for the diagnosis and care of patients with chronic obstructive pulmonary disease (COPD) and asthma. Am Rev Respir Dis 1987; 136: Collard P, Dury M, Delguste P, et al. Movement arousals and sleep-related disordered breathing in adults. Am J Respir Crit Care Med 1996; 154: Rechtschaffen A, Kales A. A manual of standardized terminology, techniques and scoring system for sleep stages of human subjects. Los Angeles, CA: University of California at Los Angeles Brain Information Service/Brain Research Institute, American Sleep Disorders Association. EEG arousals: scoring system and examples; a preliminary report from the Sleep Disorders Atlas Task Force of the American Sleep Disorders Association. Sleep 1992; 15: Hauri P. Basic facts on sleep. In: Hauri P, ed. The sleep disorders. Kalamazoo, MI: Upjohn, 1997; Vos PJE, Folgering HThM, van Herwaarden CLA. Predictors for nocturnal hypoxemia (mean Sao 2 90%) in normoxic and mildly hypoxic patients with COPD. Eur Respir J 1995; 8: Fletcher EC, Miller J, Divine GW, et al. Nocturnal oxyhemoglobin desaturation in COPD patients with arterial oxygen tensions 60 mm Hg. Chest 1987; 92: Whittle AT, Finch SP, Mortimore IL, et al. Use of home sleep studies for diagnosis of the sleep apnoea/hypopnoea syndrome. Thorax 1997; 52: Ninane V, Art G, Dachy B, et al. Polysomnography in the home [abstract]. Am J Respir Crit Care Med 1994; 149:A52 24 Leitch AG, Clancy LJ, Leggett RJE, et al. Arterial blood gas tensions, hydrogen ion, and electroencephalogram during sleep in patients with chronic ventilatory failure. Thorax 1976; 31: Fleetham J, West P, Mezon B, et al. Sleep, arousals and oxygen desaturations in chronic obstructive pulmonary disease: the effect of oxygen therapy. Am Rev Respir Dis 1982; 126: Calverly PMA, Brezinova V, Douglas NJ, et al. The effect of oxygenation on sleep quality in chronic bronchitis and emphysema. Am Rev Respir Dis 1982; 126: Arand DL, McGinty DJ, Littner MR. Respiratory patterns associated with hemoglobin desaturation during sleep in chronic obstructive pulmonary disease. Chest 1981; 80: Freedman NS, Kotzer N, Schwab RJ. Patient perception of sleep quality and etiology of sleep disruption in the intensive care unit. Am J Respir Crit Care Med 1999; 159: Chaouat A, Weitzenblum E, Krieger J, et al. Association of chronic obstructive pulmonary disease and sleep apnea syndrome. Am J Respir Crit Care Med 1995; 151: Clinical Investigations