Multiple Naps and the Evaluation of Daytime Sleepiness in Patients with Upper Airway Sleep Apnea

Transcription

1 Sleep. 3(3/4): Raven Press. New York, Multiple Naps and the Evaluation of Daytime Sleepiness in Patients with Upper Airway Sleep Apnea T. Roth, K. M. Hartse, F. Zorick, and W. Conway Sleep Disorders and Research Center, Henry Ford Hospital, Detroit, Michigan In a patient's daily life, excessive daytime sleepiness (EDS) is the most disabling symptom associated with the sleep apnea syndrome. Although the potential medical complications associated with this disorder-including right heart failure, nocturnal cardiac arrhythmias, nocturnal hypoxia, and high blood pressure-may have severe consequences for the patient's health, sleepiness in and of itself also carries a potentially lethal risk. Occupational and car accidents, for example, are not uncommon among these excessively sleepy individuals (Broughton and Ghanem, 1976; Broughton et ai., 1978) and probably increase mortality risks on a day-to-day basis by a far greater factor than do the medical complications of the disease. In addition, untreated sleepiness is very often associated with an insidious deterioration in the patient's quality of life. Loss of employment, poor school performance, and poor social relationships due to an inability of the patient to remain functionally alert and to interact with his environment are among the most bitter of complaints in individuals affected with EDS. As sleep clinicians are well aware, severe EDS is not a benign condition from either a medical or psychosocial point of view, and it is often the first symptom of underlying sleep pathology. The measurement of EDS in the sleep apnea syndrome presents a special set of problems. As Dement et al. (1978) point out, the patient's subjective estimation of his own sleepiness may be grossly inaccurate, either because he is so sleepy that he no longer has a frame of reference for what the state of alertness actually is or because he tends to deny his sleepiness for fear of being labeled as lazy or unmotivated. Subjective rating scales such as the Stanford Sleepiness Scale (SSS) which are highly correlated in normal subjects with performance and polygraphically measured sleep onset (Dement, 1976; Hoddes et al., 1973) are of little value in capturing the magnitude of the apnea patient's objective sleepiness (Dement et al., 1978). In our own experience, we have had patients literally falling asleep before our eyes, yet these patients rate themselves as fully alert on the SSS. For the evaluation of EDS patients, then, another measure of daytime sleepiness which is not dependent on the patient's subjective response is clearly needed. The Stanford group recently introduced the MUltiple Sleep Latency Test Accepted for publication September Address correspondence and reprint requests to Dr. Roth at Sleep Disorders and Research Center, Henry Ford Hospital, 2799 Grand Blvd. W., Detroit, Michigan

2 426 T. ROTH ET AL. (MSLT) as a more direct and objective means of studying daytime somnolence. In the MSLT, the patient is required to He down at several different times during the day and is given the instruction to try and fall asleep. The underlying assumption of the MSLT is that the excessively sleepy individual will have a greater tendency to fall asleep, Le., shorter latency to sleep onset, than an individual who is not pathologically sleepy. Patients who are affected with narcolepsy and upper airway apnea, sleep disorders which both have EDS as their major symptom, show significantly shorter latencies to sleep onset in comparison to normal subjects (Dement et ai., 1978; Hartse et ai., 1980c; Richardson et ai., 1978). The MSLT also provides repeated opportunities to observe multiple rapid eye movement (REM) onsets for the polygraphic diagnosis of narcolepsy (Hartse et ai., 1980a; Mitler et ai., 1979). In addition, normal subjects who are made experimentally sleepy during the day as a result of nocturnal sleep deprivation also show decreased sleep latency during the MSLT (Carskadon and Dement, 1979). All of these studies provide convincing evidence that the MSLT is an effective tool for polygraphically quantifying the patient's complaint of EDS. The purpose of this paper is to present additional data on daytime sleep tendency in a group of patients affected with upper airway sleep apnea. Two aspects of daytime somnolence will specifically be addressed. First, given that apnea patients are more sleepy than normal subjects, what feature, if any, of their nocturnal sleep or nocturnal respiratory irregularities contribute to their daytime sleepiness? Second, although the MSLT is an effective tool for discriminating normal from pathologically sleepy individuals, is it also effective in documenting, through polygraphic measurement of sleep, the dramatic subjective improvements which patients report following treatment? METHOD A sample of 10 patients, shown in Table 1, with the diagnosis of upper airway sleep apnea were selected for this report. Each patient had a several year history of EDS and/or loud snoring, and each patient completed a routine protocol which included 1 night of polysomnography followed by 1 day of multiple naps. For evaluation of sleep apnea, the nocturnal recording included standard electroencephalographic (EEG), electromyographic (EMG), and electro-oculographic (EOG) parameters, as well as measures of respiration. Airflow was measured by nasal and oral thermistors, and respiratory effort by a mercury-filled capillary strain guage stretched over the abdomen. Nocturnal oxygen saturation was continuously monitored with a Hewlett-Packard ear oximeter. A V2 or V5 electrocardiogram lead recorded heart rate. Electrodes placed over the anterior tibialis muscle recorded leg movements. On the day following the nocturnal recording, the patient was instructed at 1000, 1200, 1400, and 1600 hr to lie down on a bed in a quiet, dark room and to try and fall asleep. Each nap lasted 30 min if sleep did not occur. If stage 1 or REM appeared, the nap was terminated 15 min after the emergence of either stage. Thus, the patient could potentially spend 45 min in bed during each nap. However, due to the extreme sleepiness of these individuals, the naps typically lasted from Sleep, Vol. 3, No. 3/4, 1980

3 DA ITIME SLEEPINESS AND SLEEP APNEA 427 TABLE 1. Identifying characteristics of 10 upper airway sleep apnea patients Age Weight Chief Patient (yr) Sex (lb) Height complaint JL 36 M 287 5'9" Severe EDS KP 51 M 160 5'8" Severe EDS JG 67 M 319 5'11" Severe EDS TT 58 M 178 5'6" Severe EDS BZ 21 M 260 6'2" Severe EDS JT 58 M 235 6'4" Snoring KS 39 M 148 5'4" Mild EDS RL 50 M 235 6'0" Severe EDS AR 58 F 205 5'3" Severe EDS CM 67 M 197 5'10" Severe EDS EDS, excessive daytime sleepiness. 15 to 18 min, i.e., a 0-3 min sleep onset. The patients were instructed not to sleep between naps, and they were monitored by technicians to enforce wakefulness. However, some patients did fall asleep for brief periods despite these precautions. Ten normal, age-matched subjects without a history of EDS, snoring, or nocturnal sleep disturbance were recorded during 1 night followed by 1 day of naps to provide a comparison group for the apnea patients. Inspection of the nocturnal recordings revealed that 2 of these subjects had infrequent episodes of central apnea during the latter part of the night. Five (patients JL, BZ, JG, KS, and EC) of the 10 patients were evaluated with nocturnal polysomnography and the daytime MSLT both before and 2-5 months after treatment of their condition. One patient (BZ) was treated with 10 mg protriptyline, a tricyclic antidepressant with stimulating properties which has been shown to be efficacious in relieving both the respiratory abnormalities and the daytime somnolence associated with sleep apnea (Clark et ai., 1979; Conway et ai., 1979). The other 4 patients underwent uvulopalotopharyngoplasty, a new surgical procedure designed to increase the size of the oropharyngeal space (Fujita et al., 1980). All nocturnal polysomnographic recordings were scored for sleep stages using standard criteria (Rechtschaffen and Kales, 1968). Obstructive apneic episodes were defined by a greater than 10 sec cessation of airflow at both the nose and mouth with continued respiratory effort. Central episodes were defined by a greater than 10 sec cessation of both airflow and respiratory effort. Likewise, a greater than 10 sec criterion was imposed for the scoring of mixed episodes. Ear oximeter tracings were scored for both the total minutes and the number of times during the sleep period that percent Sa0 2 dropped below 85%. To give a more sensitive measure of arousals, other than shifts to stage 1 sleep which accompany the termination of apneic episodes, increases in chin muscle tonus and leg movements occurring with respiratory-related arousals (apneic or hypoventilation episodes) were scored "arousals subsequent to respiration" (ASR). As pointed out by Carskadon and Dement (1979), the scoring of sleep onset Sleep, Vol. 3, No. 3/4, 1980

4 428 T. ROTH ET AL. during daytime naps must be modified somewhat from the usual criteria from scoring sleep onset at night. Since the tendency to sleep rather than the amount or maintenance of sleep is the variable of interest in the MSL T, the latency to stage 1 rather than the latency to stage 2 sleep better reflects sleep tendency, particularly in apneic patients. Also, normal subjects will almost invariably reach stage 2 if they fall asleep during daytime naps. Apneic patients, on the other hand, often do not reach stage 2 due to frequent arousals related to the resumption of breathing, and therefore, their latency to sleep onset using a stage 2 criterion would be artificially long. In this report, the first 30 sec epoch of stage 1 sleep, most often scored as the disappearance of the occipital alpha rhythm rather than as the appearance of vertex waves in the central EEG, was the criterion for sleep onset in both the normal and apneic subjects. To establish interjudge scoring reliability of daytime naps, 32 naps from the apnea patients were randomly selected and scored by two independent judges. Across the 32 naps, there was an 84% agreement for epoch-by-epoch scoring between the two judges. In scoring latency to stage 1, there was perfect agreement for 47% of the naps, and for 44% of the naps, there was a Y2-1 min discrepancy. In scoring total minutes of stage 1, there was no interjudge discrepancy for 22% of the naps, and for 5990 of the naps, there was a discrepancy of Y2-2 min. These reliability figures show that daytime nap parameters, like measures of nocturnal sleep, can be reliably scored using standard Rechtschaffen and Kales (1968) procedures. Pearson r correlations were performed between nocturnal and nap parameters for the combined group of both normal subjects and apnea patients (n = 20) and individually for the normal (n = 10) and sleep apnea groups (n = 10). For the combined group and the normal group, the 4 nap mean of each of seven nap parameters (latency to stage 1, latency to stage 2, total minutes stage 1, total minutes stage 2, total minutes wake, and the SSS before and after each nap) was correlated with nocturnal sleep variables judged to have a potential effect on daytime sleepiness. These nocturnal variables included total sleep time (TST), both including and excluding stage 1; number of awakenings; percent wake; percent stage 1; and shifts to stage 1 per hour of TST. These same correlations were performed for the apneic patients, and in addition, nocturnal respiratory variables were correlated with each of the seven nap parameters. These respiratory paramaters included number of apneas per hour TST, mean duration of apneas, minutes below 85% Sao 2 per hour TST, number oftimes Sao 2 fell below 85% per hour tst, and ASR per hour TST. To test for significant differences between the group means of normal subjects and apnea patients, two-tailed t -tests were used. RESULTS As Table 2 shows, 9 of the 10 patients had severe upper airway apnea, with a mean number of episodes ranging from 28.0 to 91.0/hr. Nocturnal oxygen saturation parameters showed a wide variation, ranging from no hypoxia (patient AR) to severe hypoxia (patient JL). All patients had numerous arousals (ASR) in conjunction with respiratory irregularities. Although ASR does include some shifts to Sleep, Vol. 3, No. 3/4, 1980

5 DAYTIME SLEEPINESS AND SLEEP APNEA 429 TABLE 2. Nocturnal respiratory variables in JO upper airway sleep apnea patients No. apneas A verage duration Sa0 2 less than 85% per of apneas Patient hour (sec) Minlhr No.lhr ASRlhr JL KP JG TT BL JT KS RL AR em " No data available. ASR, arousals subsequent to respiration. stage 1 sleep, this measure is a more sensitive indicator of sleep disruption in apnea patients than are shifts to stage 1. For example, sleep apneics remain in stage 1 and frequently do not reach stage 2 sleep for long periods of time due to the almost continuous arousals associated with respiratory irregularities. A count of shifts to stage 1 does not reflect the same magnitude of nocturnal sleep disruption that a count of ASR does. In comparison to normal, age-matched subjects, the nocturnal sleep of apnea patients was significantly worse, as shown in Table 3. The apnea patients had more stage 1 (44.9 vs. 10.1% for normals) and less stage 3-4 (0 vs. 3.0% for normals). Although the difference in percent stage 3-4 was not statistically significant, this lack of difference can be attributed to the ages of the two groups. Seven of 10 subjects in both groups were over 50 years of age, and the effect of age is TABLE 3. Nocturnal sleep variables in JO upper airway sleep apnea patients and 10 normal, age-matched subjects Shifts to Stage percent stage I Latency (min) TST per hour Group (min) Wake REM TST to stage I to stage 2 Normals ± ± ± ± ± ± ± ± ± Apnea ± ± ± ± ± ± ± ± ± t value b b 5.07" 2.70" 2.07 "p < 0.02, b P < 0.01,,. p < Values are means ± SD. Sleep. Vol. 3, No. 3/4, 1980

6 430 T. ROTH ET AL. reflected in the low stage 3-4 percentage even in the normal sample. REM percent was also significantly decreased in the apnea population. Sleep onset at night, like sleep onset during the day, was more rapid in the apnea patients. The mean latency to stage 1 was 20.4 min for the normals, but only 1.9 min for the apnea patients. This difference was statistically significant (p < 0.02). Similarly, latency to stage 2 was also shorter for the apnea group, although the difference failed to reach statistical significance. The two groups also differed in the continuity of nocturnal sleep. Apnea patients had four times the number of shifts to stage 1 (per hour TST) as the normal subjects did. This difference was significant (p < 0.001), and in fact, it was the most significant difference between the two groups of any nocturnal sleep parameter. It thus appears that a characteristic feature of the apnea patient's nocturnal sleep is that of repeated arousals measured both by shifts to stage 1 and ASR. It is of interest to note that although the apnea patients had more frequent arousals than the normals, they did not have more nocturnal awakenings or an increased wake percent. The apnea patients were sleepier than normal subjects during daytime naps, as shown in Table 4. The mean latency across 4 naps to stage 1 was 12.9 min for normal subjects, whereas apnea patients had a 2.6 min latency. Similarly, the latency to stage 2 was 20.4 min for normals, but only 13.5 min for the apnea group. Both of these differences in sleep latency were significant (p < 0.001). Apnea patients also had significantly more total minutes stage 1 (12.1 vs. 6.2 min for normals) and less wake (2.9 vs. 6.5 min for normals) per nap. In addition to exhibiting a greater degree of daytime sleepiness than normals, the apnea group was more consistent across subjects in daytime sleep onset latencies. Whereas the TABLE 4. Daytime nap variables in normals and upper airway sleep apnea patients Normals Apneic patients Variable (n = 10) (n = 10) Latency to stage " ± 7.9 ± 1.7" Latency to stage " ±4.9 ±4.1 Minutes of stage " ±3.4 ±3.1 Minutes of stage ±2.2 ±2.5 Minutes of wake " ±6.5 ± 2.9" SSS before ±0.9 ±l.l a Significantly different from normals, p < b Significantly different from normals, p < Means ± SD for 4 naps are given. SSS, Stanford Sleepiness Scale. sl~ep, Vol. 3, No. 3/4, 1980

7 DA IT/ME SLEEPINESS AND SLEEP APNEA nap mean latency to stage 1 ofthe normal subjects ranged from 1.8 to 12.9 min, this value ranged from 0.3 to 5.5 min in the apnea patients. The standard deviation for latency to stage 1 was 7.9 min for normals and only 1.7 min for apnea patients. An F-test showed these standard deviations to be significantly different (p < 0.001). This finding suggests not only that there is a wide range ofpolygraphical1y defined daytime sleepiness which exists in normal subjects who do not perceive themselves as being excessively sleepy, but also that short sleep latencies are stereotyped among apnea patients. Previous studies have shown that in normal subjects, the SSS is highly correlated with the latency to polygraphic sleep onset (Dement, 1976). Dement et al. (1978), found that the SSS in EDS patients does not bear this same relationship to polygraphic sleep latency as it does in normals. In our sample, although there was a highly significant difference between normals and apneic patients in stage 1 latency, there was no significant difference between the two groups in SSS ratings preceding the naps. In other words, the apnea patients rated themselves as subjectively alert as normal subjects did. These findings stress the importance of polygraphically evaluating daytime somnolence even in the "asymptomatic" apnea patient who vigorously denies EDS. It is clear from these findings that apnea patients have more disrupted nocturnal sleep and that their daytime somnolence is of a greater magnitude than that of normal subjects. The question which arises is whether any parameter of nocturnal sleep is correlated with daytime sleep tendency in these two groups. Significant correlations between nocturnal sleep variables and daytime nap parameters are shown in Table 5. For the combined group (n = 20) of apneic and normal subjects, the nocturnal parameters which significantly and consistently correlated with daytime nap parameters were shifts to stage 1 and percent stage I per hour TST. The more frequently the subject aroused during the night (number of shifts to stage 1 and total percent stage 1), the more somnolent the subject was during daytime nap recordings (shorter latencies to stages 1 and 2 and fewer epochs of wake). Simply stated, these correlations show that the greater the disruption of nocturnal sleep, the more likely it is that sleep will occur the following day. To determine if this relationship existed in each of the normal and apnea populations, separate correlations were carried out for the two groups. The results showed that the relationship between shifts to stage 1 and daytime somnolence was stronger in normal subjects than the total group, and that this same relationship did not exist in the apnea patients. In normals, then, we can conclude that daytime somnolence is best predicted by nocturnal arousals (shifts to stage 1). A somewhat unexpected finding was that nocturnal percent wake was positively correlated with three nap variables and that, in addition, it was negatively correlated with total minutes of stage 1 sleep during daytime naps. In other words, subjects who had the greatest amount of wakefulness at night also had the longest sleep latency, the most wakefulness, and the fewest minutes of stage 1 sleep during daytime naps. These findings suggest that although nocturnal sleep disruption does affect daytime sleepiness, the latency to sleep onset during daytime naps may be to some degree "set" in normal individuals by a base-line level of alertness such that subjects with a "high alertness" level, reflected by an increased Sleep. Vol. 3, No. 3/4, 1980

8 432 T. ROTHET'AL TABLE 5. Significant Pearson r Correlations between nocturnal and nap variables Nocturnal sleep parameters Nap parameters (Mean of naps 1-4) Total Sample (n = 20) Apneic patients Normals (n = 10) (n = 10) Latency to stage 1 Latency to stage 2 Total minutes wake Total minutes stage 1 SSS before Shifts to stage 1 per hr TST Stage 1 percent Shifts to stage 1 per hr TST Shifts to stage 1 per hr TST Stage 1 percent ASR, arousals subsequent to respiration. Shifts to stage 1 per hr TST Wake percent 0.66 Shifts to stage 1 ASR per hr per hr TST Apneas per hr Wake percent Stage 1 percent Shifts to stage I per hr TST Wake percent 0.73 Wake percent wake percentage at night, may also have a long latency to sleep onset during daytime naps. Unlike the normal subjects, no nocturnal sleep parameter significantly correlated with any parameter in the daytime naps of apnea patients. This suggests that nocturnal sleep disruption, as measured by conventional criteria, is not a contributing factor in the apnea patients' daytime somnolence. However, it is important to keep in mind that the absence of significant variability in some nap p~ameters-for example the latency to stage 1 sleep, which was very short in au apnea patients-may account for this lack of correlation. Nocturnal respiratory variables (minutes below 85% Sao2 per hour TST, number of times below 85% Sao2 per hour TST, and length of apneas, number of apneas per hour TST) did not contribute significantly to daytime sleep tendency. Only ASR per hour of TST had a significant negative correlation with stage 2 latency in the naps; the greater the ASR, the shorter the latency to stage 2. Although the correlation was not significant (r = -0.52), the number of apneas per hour TST was the only other respiratory measure which had a relatively high correlation with either stage 1 or stage 2 latency in naps. In summary, the apnea patients, like the normal subjects, showed a relationship between daytime sleep latency and nocturnal sleep disruption. In the case of normal subjects, shifts to stage 1 showed the highest correlation with both stage 1 Sleep, Vol, 3, No. 3/4, 1980

9 DAYTIME SLEEPINESS AND SLEEP APNEA 433 and stage 2 latency, whereas in apnea patients, ASR, rather than shifts to stage 1, showed a significant correlation only with stage 2 latency. To further investigate the relationship between nocturnal and daytime sleep of apnea patients, 5 patients were compared before and after treatment. Before treatment, all complained of EDS, all had significant upper airway sleep apnea (mean, 66.5 apneas/hr), and all showed objective evidence of daytime somnolence (mean latency to stage 1 sleep, 3.3 min). Clinically, all patients except one (JG) reported an improvement in their daytime sleepiness following treatment. Examination of nocturnal sleep parameters showed that the sleep of these patients did, in fact, improve with treatment. As shown in Table 6, all 5 had less disturbed nocturnal sleep post- as compared to pretreatment. Shifts to stage 1 per hour TST were significantly (p < 0.02) reduced from a mean of 14 to 5, and percent stage 1 was significantly (p < 0.001) reduced from 48 to 32%. Consistent with this improvement in nocturnal sleep, nocturnal respiratory parameters (Table 7) also improved. Arousals associated with respiration (ASR) were significantly (p < 0.02) decreased from 43.1 to 15.6/hr). Although statistical significance was not obtained, the number of apneas was also reduced in each of the 5 patients, showing a decline from a group mean of 66.5 to 38.2/hr. Based on the improvement in nocturnal respiratory variables, the reduction of nocturnal sleep disruption, and the patient's own reports, it was anticipated that there would be a reduction in polygraphically defined daytime somnolence following treatment. However, there were no striking changes in objective measures of daytime sleep tendency. Although 4 of the 5 patients reported subjective improvements posttreatment, only 1 had an increased stage 1 latency (patient JL), and the other 4 patients actually had a slightly shorter latency to sleep onset following treatment. The only variable which approached a statistically significant posttreatment change was total minutes of stage 1 sleep. Four ofthe 5 patients had less stage 1 posttreatment compared with pretreatment. DISCUSSION The major findings in this study were that nocturnal sleep in apnea patients is more disrupted than that of normal subjects, that daytime somnolence is more pronounced in apnea patients than in normal subjects, and that polygraphic measures of daytime somnolence in both groups are highly correlated with nocturnal sleep disruption. However, improvement in the nocturnal sleep of apnea patients was not accompanied by an improvement in polygraphically defined measures of sleepiness, as shown by the results of pre- and posttreatment evaluations in 5 patients. There has been much speculation concerning the underlying cause of EDS in upper airway sleep apnea. Given that nocturnal sleep deprivation in normal subjects produces rapid sleep onset during daytime naps (Carskadon and Dement, 1979), it might reasonably be assumed that respiratory-induced sleep deprivation in apnea patients is also related to EDS in this population. This relationship between reduced nocturnal sleep and EDS, however, was not apparent in the Sleep, Vol. 3, No. 314, 1980

10 '" ~ " ~,'-- ~ '" ~ TABLE 6. Pre- and posttreatment nocturnal sleep variables in 5 upper airway apnea patients Wake Min TST Shifts to stage 11 Min TST less stage 1 hr TST Number Percent Stage 1% ~ Patient Pre Post Pre Post Pre Post Pre Post Pre Post Pre Post ::0 \) JL ~ BZ trl JO "'"l KS > EC t""-< Mean t Value Probability ns <0.01 <0.02 ns ns <0.001 ~ ~

11 DAYTIME SLEEPINESS AND SLEEP APNEA 435 TABLE 7. Pre- and posttreatment nocturnal respiratory variables in 5 upper airway apnea patients Apnea SaO. less than 85% Mean No'/hr TST length (sec) Minlhr TST No./hr TST ASRlhr TST Patient Pre Post Pre Post Pre Post Pre Post Pre Post JL BZ JG KS EC Mean t Value Probability ns ns ns <0.05 <0.02 ASR, arousals subsequent to respiration. TST, total sleep time. present data. Total sleep time and percentage wake were not significantly different between our two populations. Since individuals affected with sleep apnea are not able to sustain wakefulness during the day, it is not surprising that they would, in tum, be able to sustain wakefulness during the nocturnal sleep period following an arousal. Therefore, these measures, which are commonly used to evaluate the "goodness" of nocturnal sleep in normal subjects, are not sufficient to evaluate the quality of nocturnal sleep in apnea patients. Similarly, in the "asymptomatic" patient without a complaint of EDS, a recent report found that the difference in apnea patients who complain (symptomatic) and those who do not (asymptomatic) complain of EDS cannot be accounted for on the grounds that sleep is more disturbed in the symptomatic group (Orr et ai., 1979). However, these investigators used the criterion of percent wake and movement time during nocturnal sleep as the basis for their conclusion. In our TABLE 8. Pre- and posttreatment nap variables in 5 upper airway apnea patients Latency (min) to stage 1 to stage 2 Wake (min) Stage 1 (min) SSS Patient Pre Post Pre Post Pre Post Pre Post Pre Post JL HZ JG KS EC Mean t Value Mean values for 4 naps are given. SSS, Stanford Sleepiness Scale. Sleep, Vol. 3, No. 314, 1980

12 436 T. ROTH ET AL. data, at least one of these parameters-percent wake-does not even distinguish normal subjects from apnea patients, yet the apnea patients are objectively more sleepy during daytime naps. We have shown here that the brief arousals following the end of an apnea episode, which in most instances do not lead to a stage change or sustained wakefulness, are most highly correlated with stage 2 latency during daytime naps, and in normals, shifts to stage 1 correlated with both stage 1 and stage 2 latency during naps. However, it is important to keep in mind that stage 1 latency, which is used as the measure of daytime sleep tendency, did not correlate with respiratory-related arousals in the apnea patients. Therefore, it is still not clear from these data that arousals from sleep rather than the amount of sleep predict tendency to sleep during the day. None of the nocturnal respiratory parameters which related to either number and duration of apneas or to oxygen saturation were significantly correlated with daytime somnolence, although the mean number of apneas per hour showed a trend in this direction. Again, it is possible that the parameters which we chose to evaluate may not be those which are most critical to the manifestation of daytime sleepiness. For example, the failure of minutes below 85% Sao2 and the number of times Sao 2 fell below 85% to correlate with stage 1 nap latency could be due to the fact that 85% is too low or high a criterion. A different criterion may be more sensitive in predicting daytime stage 1 sleep latency. A major difference between our data and those of a previous study (Dement et al., 1978) lies in the sensitivity of daytime nap tests for polygraphic validation of subjective improvement in apnea patients following treatment. However, these differences may be explained by the types of treatments which the patients in the two studies received. In the study of Dement et al. (1978), every patient received a tracheostomy for treatment of his condition, and every patient had a dramatic decrease in polygraphic daytime somnolence. Although no figures are given, it must be assumed that apneic episodes were completely eliminated in these patients, since tracheostomy bypasses the point of upper airway obstruction. Only one (patient BZ) of the 5 patients for whom we present data had, for practical purposes, no apnea (2 apneas/hr) or hypoxia following treatment. All other patients still had a moderate degree of both apnea and hypoxia following treatment. In comparison to Dement et al., then, patients in our sample were still affected with apnea, and the lack of objective improvement in daytime somnolence reflects this fact. It is also possible that our patients subjectively overestimated their improvement following treatment. A more general issue can be raised regarding the use of the MSLT as a tool for the polygraphic diagnosis of EDS. There is an unambiguous difference between the tendency of normal subjects and the tendency of individuals affected with sleep pathologies such as apnea and narcolepsy to sleep in daytime naps (Dement, et ai., 1978; Richardson et ai., 1978; Hartse et ai., 1980c), and there is no question that the MSLT effectively discriminates between pathological and normal sleepiness. However, as a measure of subtle variations in sleepiness, the MSLT may not be appropriate. For example, the short sleep latencies following treatment in our subjects clearly do not parallel their subjective improvement, their improved Sleep, Vol. 3, No. 314, 1980

13 DAYTIME SLEEPINESS AND SLEEP APNEA 437 nocturnal sleep, or their improved nocturnal respiration. Furthermore, we have also studied a group of 5 narcoleptic patients both before and after treatment. All patients reported dramatic changes in their ability to sustain wakefulness posttreatment. Howwever, there was no change in daytime sleepiness. The mean latency across all naps and all subjects to stage 1 sleep pretreatment was 2.78 ± 2.07 min, and the mean latency posttreatment was 1.75 ± 2.07 min. It may be questioned, then, whether the posttreatment change which we wish to document is a decrease in the tendency to fall asleep as measured by the MSLT or whether a more important variable to consider would be a posttreatment improvement in the ability to sustain wakefulness. One possible method of evaluating changes in the ability to sustain wakefulness is through daytime performance testing. However, this approach presents difficulties, since many EDS patients fall asleep under the monotonous conditions of performance tasks and require that the experimenter continuously arouse the subject if any "performance" is to be elicited. Another approach to this problem is to test directly the patient's ability to remain awake, that is, instead of instructing the patient to fall asleep during daytime naps, ask the patient to remain awake. If the patient's pretreatment complaint is that he cannot remain awake, then an instruction to remain awake should have little effect on his tendency to fall asleep. If, on the other hand, the patient's posttreatment report is that he is now able to sustain wakefulness, then this increased "wakefulness ability" should be demonstrated by longer sleep latencies when the patient is given an instruction to remain awake. To determine if instruction changes the latency to sleep onset in normal subjects, we recently conducted a study in which subjects without a history of EDS, snoring, or nocturnal sleep disturance were given "stay awake" and "fall asleep" nap instructions counterbalanced with nights 1 and 2 in the laboratory (Hartse et ai., 1980b). The results showed that nap instructions did have a significant (p <0.01) effect on sleep latency. The mean stage 1 sleep latency in the "fall asleep" condition was 12.2 ± 2.2 min which increased to 17.0 ± 2.4 min in the "stay awake" condition. Similarly, the mean stage 2 latency in the "fall asleep" condition was 19.3 ± 1.9, min and increasing to 24.0 ± 2.1 min in the "stay awake" condition. These findings indicate that in normal subjects at least, sleep latency can be manipulated by verbal instruction. Although the mean latency to both stages 1 and 2 increased only by about 5 min in the "stay awake" condition, it should be kept in mind that in both conditions the subjects were lying down with theileyes closed in a quiet, dark room. The MSLT in its present format is a valuable tool primarily for diagnosis of the presence or absence of polygraphically defined EDS. However, as new surgical and pharmacological treatments for apnea are developed which have as their primary goal the reduction rather than the complete elimination of apneic episodes, new procedures should be explored to sensitively quantify the degree of EDS which is manageable from the patient's point of view on a day-to-day basis. Successful treatment of the excessive somnolence associated with upper airway apnea, then, might be more realistically judged in terms of the increased propensity to maintain wakefulness rather than the decreased propensity to fall asleep. Sleep, Vol. 3, No. 3/4, 1980

14 438 T. ROTH ET AL. REFERENCES Broughton R and Ghanem Q. The impact of compound narcolepsy on the life of the patient. In: C Guilleminault, WC Dement, and P Passouant (Eds), Narcolepsy, Spectrum, New York, 1976, pp Broughton R, Nevsimalova S, and Roth B. The socioeconomic effects (including work, education, recreation and accidents) of idiopathic hypersomnia. Sleep Res 7:217, Carskadon MA and Dement WC. Effects of total sleep loss on sleep tendency. Percept Mot Skills 48: , Clark RW, Schmidt HS, Schaal SF, Boudoulas H, and Schuller DE. Sleep apnea: Treatment with protriptyline. Neurology 29: , Conway W, Zorick F, Hartse KM, Piccione P, and Roth T. Protriptyline in the treatment of sleep apnea. Sleep Res 8:177, Dement WC. Daytime sleepiness and sleep "attacks." In: C Guilleminault, WC Dement, and P Passouant (Eds), Narcolepsy, Spectrum, New York, 1976, pp Dement WC, Carskadon MA, and Richardson G. Excessive daytime sleepiness in the sleep apnea syndrome. In: C Guilleminault and WC Dement (Eds), Sleep Apnea Syndromes, Alan R Liss, New York, 1978, pp Fujita S, Zorick F, Conway W, Roth T, Hartse KM, and Piccione P. Uvulo-palato-pharyngoplasty: A new surgical treatment for upper airway sleep apnea. Sleep Res 9: 1980 (in press). Hartse KM, Roth T, Zorick F, and Moyles TP. REM sleep episodes during multiple daytime naps of narcoleptic subjects. Sleep Res 9: 1980a (in press). Hartse KM, Roth T, Zorick FJ, and Zammit G. The effect of instruction upon sleep latency during multiple daytime naps of normal subjects. Sleep Res 9: 1980b (in press). Hartse KM, Zorick FJ, Roth T, Kaffeman ME, and Moyles TP. Daytime sleep tendency in normal, insomniac and EDS populations. Sleep Res 9: 1980c (in press). Hoddes E, Zarcone V, Smythe H, Phillips R, and Dement W. Quantification of sleepiness: A new approach. Psychophysiology 10: , Mitier MM, Van den Hoed J, Carskadon MA, Richardson G, Park R, Guilleminault C, and Dement WC. REM sleep episodes during the multiple sleep latency test in narcoleptic patients. Electroencephalogr Clin Neurophysiol 46: , Orr WC, Martin RJ, Imes WK, Rogers RM, and Stahl ML. Hypersomnolent and nonhypersomnolent patients with upper airway obstruction during sleep. Chest 75: , Rechtschaffen A and Kales A. A Manual 0/ Standardized Terminology, Techniques and Scoring System/or Sleep Stages 0/ Human Subjects. Brain Information Service/Brain Research Institute, University of California at Los Angeles, Richardson GS, Carskadon MA, Flagg W, Van den Hoed J, Dement WC, and Mitler MM. Excessive daytime sleepiness in man: Multiple sleep latency measurement in narcoleptic and control subjects. Electroencephalogr Clin Neurophysiol 45: , Discussion Dr. Weitzman commented that this work touched on a fundamental question about sleeping and waking; viz., are both processes active and not mutually exclusive, or is one due to the absence of the other? The first idea means that elements of each state can occur together. Dr. Roth said that their data suggested that both states can go on independently and when you improve the obstructive apnea it is not the sleepiness so much as the tendency to stay awake that is improved. Dr. Phillipson noted that work of Orr et a1. (1979) (see above) suggested that the oxyhemoglobin saturation was the crucial factor in excessive daytime sleepiness (EDS). Dr. Guilleminault said they thought that sleep fragmentation was the important factor and that hypoxemia could not be the only answer. He quoted work of Dr. Weil's group in which normals, subjected to nasal obstruction, woke up before any oxyhemoglobin desaturation occurred. He said that the lack of an objective measure of alertness was a major Sleep, Vol. 3, No. 3/4, 1980

15 DAYTIME SLEEPINESS AND SLEEP APNEA 439 problem in studying this area. He noted that Dr. Roth's data did not show a large difference in the MSLT. Dr. Roth replied that there was a statistically significant difference in their MSL T data and noted that there was a dramatic change in some patients, whereas some patients had no change after tracheostomy. He also said that they have some patients who report a subjective improvement but without changes in the MSLT. He noted that the data ofoit et al. (1979) were based on the patient's subjective complaint of EDS. Sleep, Vol. 3, No. 314, 1980