go to content

THE RELATIONSHIP BETWEEN SUBJECTIVE SLEEP ESTIMATION
AND OBJECTIVE SLEEP VARIABLES IN DEPRESSED PATIENTS

Vadim S. Rotenberg   vadir@post.tau.ac.il, P. Indursky, L. Kayumov, P. Sirota, Y. Melamed

INTERNATIONAL JOURNALOF PSYCHOPHYSIOLOGY www.elsevier.nl/locate/ijpsycho

Sackler School of Medicine, Tel-Aviv University, The Y. Abarbanel Mental Health Center, 15 Keren Kayemet St., Bat-Yam 59100, Israel

Received 23 September 1998; received in revised form 9 March 2000; accepted 30 March 2000

Abstract

Introduction: To our knowledge there is no evidence in the literature about the relationship between subjective sleep estimation and objective sleep variables in depression. It is not known whether the subjective estimation of sleep quality and sleep duration is directly related to any objective sleep variable in depressed patients.
Methods: Thirty patients with major depression and 10 healthy subjects have been investigated in our sleep laboratory during 1 or 2 consecutive nights after 1 night for adaptation. Every subject, after final awakening in the laboratory, answered questions concerning the subjective feelings about sleep duration, number of awakenings and sleep depth. We compared the sleep estimation in both groups and calculated the correlation between objective and subjective sleep variables in depressed patients.
Results: The degree of a wrong sleep estimation in depressed patients is larger than in healthy subjects. Slow wave sleep (SWS) in depressed patients correlates positively with the subjective estimation of sleep duration. Eye movement density in REM sleep correlates with the subjective estimation of the number of awakenings.
Conclusion: SWS in depression has a positive influence on the subjective feeling of sleep duration while phasic REM sleep activity has a negative influence. 
Keywords:   Sleep variables; Subjective estimation; Depression; Correlations

1. Introduction

Sleep in depression is investigated in more detail than sleep in any other psychiatric disorder. Slow wave sleep (SWS) deficiency, low sleep duration caused by the increased sleep latency, awakenings during the night and early morning awakenings, redidtribution of REM sleep with its concentration in the first half of the night, decreased REM sleep latency, increased eye movement (EM) density in the first cycle, absence of the first night effect . all these features characterize the sleep of depressed patients  (Kupfer et al., 1980,1988; Ansseau et al, 1985; Reynolds and Kupfer, 1987; Benson and Zarcone, 1993; Rotenberg et al., 1997a).

Although no one of the above mentioned alterations of sleep architecture is specific for depression (Benca et al., 1992), their combination displays a pattern which is mostly typical for depressed patients. On the other hand, subjective complaints on sleep disorders are very common in depression (Buysse et al., 1989), and it is very possible that they are directly related to the alterations of sleep structure. However, although in some investigations objective and subjective sleep variables are evaluated simultaneously (Parrino et al., 1994; Pasternak et al., 1994), the correlation between these two dimensions was not performed and no attempts have been made to find any specific relationships between them. In those investigations where different forms of antidepressive treatment have been used (Dietzel et al., 1986; Paiva et al., 1988; Parrino et al., 1994) it is difficult to distinguish the effect of sleep change from the effect of mood change or from the direct effect of medication on the subjective sleep variables. Only a few investigations have shown that the alterations of the objective sleep variables in depressed patients are in parallel with the subjective estimation of the same variables (like number of awakenings, sleep onset latency and total sleep time) (Hemmeter et al., 1995). However, in other investigations EEG sleep measures did not correlate with subjective sleep quality (Pasternak et al., 1992), and in double-blind placebo-maintenance therapy, an improvement in subjective sleep quality has been achieved despite the stability of most polysomnographic measures (Lee et al., 1993).

Thus, until now it is not known whether the subjective estimation of sleep quality and sleep duration in depressed patients is directly related to any (corresponding or non-corresponding) objective sleep variables and whether it is possible to predict the specific alteration of sleep structure according to the sleep complaints.

The present study was aimed at investigating the relationship between subjective and objective sleep variables in depression. According to the investigations performed on healthy subjects (Keklund and Akerstedt, 1997), and our previous investigation on insomniac patients (Rotenberg, 1982, 1993), we predicted that the total amount of SWS (Hypothesis 1) or SWS duration in particular cycles (Hypothesis 2) determine the subjective estimation of sleep depth or sleep duration.

Based on the results of our recent investigation on schizophrenic patients (Rotenberg et al., 1999) we also predicted that the subjective estimation of awakenings during the night might be related to the total EM density in REM sleep (Hypothesis 3) or EM density in particular cycles (Hypothesis 4).

Another aim of the present investigation was to compare the subjective estimation of sleep (sleep duration, sleep latency, awakenings during sleep) in depressed patients and in healthy subjects. Our prediction was that depressed patients display underestimation of sleep duration (Hypothesis 5) and overestimation of sleep delay (Hypothesis 6) and awakenings (Hypothesis 7) in comparison to healthy subjects.

2. Material and method

2.1. Study participants

We investigated 30 patients with major depression, 14 males and 16 females, with a mean age of 51.8 13.6 years. The duration of the disorder was 6 months to 17 years. The clinical diagnosis of Major Depression was based on DSM-IV criteria. The Hamilton rating scale for depression (Hamilton, 1960) was used 1 day before the first polysomnography in order to score the severity of the present depressive episode. The mean score for depression in all patients was 33.4 7.5. The mean number of the previous hospitalizations was four. All patients were physically healthy and no medical condition was responsible for their depression. In a clinical interview we have excluded other diagnoses associated with altered sleep, e.g. alcoholism, sleep apnea or myoclonus.

The absence of sleep apnea and myoclonus was confirmed in the subsequent sleep investigation. Patients were free from psychotropic medication for at least 10 days before the investigation and they were not previously treated with neuroleptics and other medications with a long half-life.

All patients gave informed consent to participate in the study, which has been accepted by the local Helsinki Committe. Sleep data were collected from all patients on the first night in the sleep laboratory, after the adaptation night. In addition, 14 patients were investigated on the second night. Lights were usually turned off at 22.00 h and turned on at 06.30 h, according to the habitual sleep/wake schedule in the ward. During the day patients were under the supervision of the staff in order to exclude napping.

We have also investigated a control group of healthy subjects (six women and four men, mean age 43.3 10.0 years) without any complaints on sleep disorders.

2.2. Measures

Frontal electroencephalogram, submental elec-tromyogram, and electrooculogram registration were performed using an electroencephalograph Neurofax EEG-4400. EEG silver electrodes were placed on the F4 position. Two channels were used for eye movement recording. One electrode was fixed almost 1 cm above the outer canthus of one eye, the second electrode was fixed almost 1 cm below the outer canthus of the other eye. Both electrodes were referred to the reference electrode on the left mastoid. EEG and EOG were recorded at a time constant of 0.3 s, EMG was recorded at a time constant of 0.03 s. The high filter setting for EEG was 70 Hz, for EOG 15 Hz, and for EMG 120 Hz. Polysomnograms (PSG) were analyzed according to international criteria (Rechtschaffen and Kales, 1968). The investigator who analyzed PSG was blind with respect to the subjects. Sleep onset was defined as the first 30-s epoch of stage I or any other sleep stage which appeared after wakefulness and was followed with no more than 1 min of wakefulness during the first 10 min of sleep. REM latency was defined as the time from the onset of sleep until the onset of the first REM sleep period. The first sleep cycle was defined as sleep duration from the sleep onset until the end of the first REM sleep period. The subsequent cycles were defined as sleep duration from the end of the preceding REM sleep period until the end of the subsequent REM sleep period.

REM percentage and SWS percentage were the percentage of total sleep time spent in REM sleep and in SWS. Eye movements (EM) in REM sleep were counted visually. The criteria for EM detection was a minimum of 25 mV (Benson and Zarcone, 1993). EM density was counted as EM frequency per minute of stage REM. Fragments of wakefulness or non-REM sleep incorporated in a single REM sleep period were not analyzed for EM activity and were not included in the REM period length when we computed a net EM/REM period variability. A minimum of REM sleep duration was defined to be more than 1 min.

2.3. Procedure

Every morning, after the final awakening in the laboratory, every subject was requested to answer questions concerning the subjective feelings about his/her sleep duration, sleep latency, number of awakenings and sleep depth (Rotenberg et al., 1997b, 1999 Appendix A). Normal group and depressed patients were compared according to the percentage of the correct estimation (error is < 30 min), overestimation and underestimation of sleep duration; according to the percentage of the correct estimation (error is < 10 min), over-estimation and underestimation of sleep delay, and according to the number of awakenings in healthy subjects and in depressed patients.

2.4. Statistics

Student's f-test was used in order to compare objective sleep variables and subjective sleep estimation in depressed patients and healthy subjects.

Pearson product-moment correlation and the Spearman correlation coefficient were used for the estimation of relationships between the subjective sleep estimation and objective sleep variables in depressed patients. In order to perform such correlations, categorical sleep ratings have been ranged. According to our initial hypotheses, we have correlated (a) the subjective estimation of sleep duration/sleep depth to the total amount of SWS (SWS%) as well as to the SWS duration in different sleep cycles; and (b) the subjective estimation of the number of awakenings to the total EM scores and EM scores in different cycles. There are no relationships between hypotheses related to SWS and EM density. We have also correlated subjective sleep duration to objective sleep duration, subjective sleep latency to objective sleep latency and subjective number of awakenings to objective number of awakenings. Only significant correlations (P < 0.05) are presented and discussed. Because correlations performed according to the first four hypotheses contain a large number of variables, we have used the Bonferroni correction for P values. These hypotheses contained five objective variables in each, thus for this analysis P < 0.01.

Table 1. Sleep variables in healthy subjects and in depressed patients.

Groups

Healthy

subjects

Depressed

subjects

Nights

20

68

Total sleep time (min)

412 (43)

347 (100)

Sleep efficiency (%)

0.98 (0.3)

0.89 (0.14)*

Sleep latency (min)

3.3 (2.1)

26 (48)*

Slow wave sleep (%)

14.5 (6.8)

10.9 (6.8)

SWS, 1 cycle (min)

26.2 (12.3)

14 (30)

SWS, 2 cycle (min)

22.6 (13.2)

10 (15)

SWS, 3 cycle (min)

6.6 (8.7)

9(18)

SWS, 4 cycle (min)

4.8 (8.6)

4.8 (7.3)

REM sleep latency (min)

89.4 (33.7)

78 (72)

REM sleep (%)

19.3 (3.7)

23 (5.6)

REM sleep 1 cycle (min)

7.9 (4.5)

22 (18)*

REM sleep 2 cycle (min)

19.7 (9.5)

22 (17)

REM sleep 3 cycle (min)

22.4(11.5)

19 (17)

REM sleep 4 cycle (min)

23.0 (12.3)

19(13)

EM density, 1 cycle (min)

2.0 (2.3)

6 (4)***

EM density, 2 cycle (min)

3.8(1.9)

5.5 (3.8)**

EM density, 3 cycle (min)

5.1 (2.6)

5.8 (5.6)

EM density, 4 cycle (min)

5.4 (3.2)

6.4 (5.3)

Awakening

1.0 (1.0)

3(3)

*P < 0.05;     **P < 0.02;     ***P< 0.01.

3. Results

The mean data of the sleep structure in depressed patients and healthy subjects are represented in Table 1. As could be expected from numerous investigations (see Section 1), depressed patients display shorter sleep duration and sleep efficiency, longer sleep latency, increased stage I, decreased stage II and SWS, increased REM sleep duration in the first cycle, increased eye movement density in the first two cycles, and increased wakefulness.

The correct estimation of sleep duration was present in 24% of all nights in depressed patients and in 25% of all nights in healthy control subjects. The underestimation was present in 42 and 55% of all nights, respectively. Differences between groups are not significant. There was a non-significant tendency for overestimation of sleep duration in depressed patients (after 33% of all nights vs. 20% of all nights in healthy control subjects). However, the mean degree of the over-estimation of sleep duration was larger in patients (88.6 32.7 vs. 50.0 11.5 min, P < 0.001), as was the degree of underestimation (-201.4143.1 vs. -90.3 46.2, P<0.01).

The correct estimation of sleep latency was present in 21.8% of all nights in depressed patients and in 30% of all nights in healthy subjects (difference is non-significant). The underestimation of sleep latency was absent in healthy subjects and present in 12% of all nights in depressed patients. The overestimation of sleep latency was present in 70% of all nights in healthy subjects and 65% in depressed patients.

The average degree of the overestimation of sleep latency was equal in both groups (42.9 35.5 min in healthy subjects vs. 46.9 28.4 min in depressed patients).

The correct estimation of the number of awakenings was present in 10% of all nights in healthy subjects and in 13% of all nights in depressed patients. The underestimation of the number of awakenings was present in 41% of all nights in depressed patients and in 10% of nights in healthy subjects (P < 0.001). The overestimation of the number of awakenings was present in 45% of all nights in depressed patients and in 78% of all nights in healthy subjects (P < 0.02). The average degree of the overestimation of the number of awakenings is almost equal in both groups (2.2 in depressed patients, 2.1 in healthy subjects). The mean degree of the underestimation of awakenings is more prominent in depressed patients (3.8 4.2 vs. 1.5 0.5, P< 0.051).

Objective sleep variables correlated only with the subjective estimation of sleep duration and with the subjective estimation of the number of awakenings. However, these subjective variables were not determined by the corresponding objective sleep variables (real sleep duration and the number of awakenings). Subjective estimation of sleep duration correlated positively with SWS% (Spearman rho = 0.370, P< 0.004, Pearson r = 0.4, P< 0.002) and with SWS duration in the second cycle (Spearman rho = 0.306, P < 0.0019, Pearson r = 0.315, P < 0.01). Subjective estimation of the number of awakenings correlated with the total EM scores (Spearman rho = 0.277, P < 0.04, Pearson r = 0.387, P < 0.005) and with the EM density in the second cycle (Spearman rho = 0.384, P< 0.005, Pearson r = 0.387, P< 0.005). Subjective estimation of sleep latency correlated with EM density in the first cycle (Spearman rho = 0.320, P< 0.021; Pearson r = 0.386, P < 0.005). No one objective variable correlated with the subjective sleep depth.

In healthy subjects we have found a significant correlation of the subjective estimation of sleep duration with SWS in the second cycle (Spearman rho = 0.62, P < 0.003).

4. Discussion

Our prediction that depressed patients, in comparison to healthy subjects, display more prominent underestimation of sleep duration was confirmed only partly. Both groups have an almost identical percentage of correct estimations of sleep duration. Moreover, depressed patients display a tendency to the more frequent overestimation of sleep duration and to the slight decrease of the number of its underestimation, in comparison to healthy subjects. However, when depressed patients overestimate and especially underestimate sleep duration the degree of this wrong sleep estimation is significantly higher than in the control group. The underestimation has a tendency to appear slightly more often than the overestimation (42 vs. 32%), and it is reasonable to suggest that the prominent underestimation may contribute in complaints on sleep duration. It may be an explanation of the more often complaints on sleep duration in depressed patients.

According to sleep latency, we have not found a significant difference between depressed patients and healthy control subjects. The over-estimation of sleep latency is equal in both groups according to the percentage of nights as well as according to the degree of overestimation. Surprisingly, there was a slight tendency to underestimate the sleep delay in depressed patients.

Our prediction according to the estimation of the number of awakenings was also not confirmed. Depressed subjects rarely overestimate and more often underestimate the number of awakenings in comparison to healthy subjects. The degree of the underestimation of awakenings is also more prominent in depressed patients. This unexpected paradoxical underestimation of awakenings may relate directly to the prominent underestimation of sleep duration, that is, to the overestimation of wakefulness during night. If a subject feels that (s)he is in a state of wakefulness being actually sleeping, than (s)he will estimate the awakening as the continuation of previous 'wakefulness', thus underestimating the number of awakenings. In our previous investigation (Rotenberg, 1982, 1993) we have shown that even healthy subjects often do not realize sleep after awakenings in NREM sleep before the REM sleep period. In insomniac patients it happens also periodically after REM sleep, suggesting that in patients REM sleep is functionally inefficient. If one does not realize the previous sleep, one is also unable to realize the awakening.

Our initial hypothesis according to relationship between subjective sleep estimation and objective sleep variables was confirmed only partly. SWS correlates positively with the subjective estimation of sleep duration but not with the subjective sleep depth and sleep quality, in contrast to healthy subjects (Keklund and Akerstedt, 1997) and to insomniac patients. (Rotenberg, 1993). The lack of SWS in depressed patients may determine the massive underestimation of sleep duration mentioned previously. It is possible to suggest that in depression other factors, like phasic REMsleep activity, may interfere with the estimation of sleep depth. In healthy subjects EM density correlates with psychic activity in REM sleep . with dream reports (Rotenberg, 1988; Hong et al., 1997). At the same time, in patients with mental disorders, the number of dream reports in REM sleep are reduced (Kramer and Roth, 1973; Rotenberg, 1993). According to the results of the present investigation, it is possible to suggest that in depression, psychic activity in REM sleep is often not perceived as dream mentation, but is considered subjectively as wakefulness. It can explain the positive correlation between eye movement density and the subjective estimation of the number of awakenings. Thus, it may be one of the reasons of the underestimation of sleep duration in depression. We do not have a (same story as above) definite explanation for the correlations between subjective sleep estimation and sleep variables in the second cycle. However, in our previous investigations (Rotenberg, 1982) we have shown that sleep variables in the second cycle are especially sensitive to the examination stress and to the level of adaptation to the shift work. Thus, this cycle may be functionally different from other cycles. However, this assumption requires further investigation.

It is reasonable to stress that the subjective sleep variables do not correlate with the corresponding objective sleep variables in depression: subjective sleep duration does not correlate with objective sleep duration, likewise subjective sleep delay does not correlate with sleep latency or the subjective number of awakenings with the number of awakenings. It may explain the lack of positive findings according to objective-subjective relationships in depression in some previous investigations.

In conclusion, it is necessary to stress that it is only a pilot investigation with some limitations: it would be reasonable to compare subjective -objective relationships in patients of different gender and age. For such investigations it would be necessary to increase the group of patients. The present data suggest that in depressed patients SWS has a positive influence on the subjective feeling of sleep duration while phasic REM sleep activity has a negative influence on the subjective sleep evaluation.

Appendix A: Sleep questionnaire (for presentation after the morning awakenings)

1.   How long have you slept?

2.   More than 9 h; 8.5 h; 8 h; 7.5 h; 7 h; 6.5 h; 6 h; 5.5 h; 5 h; < 5 h; no sleep at all.

3.   How long did it take you to fall asleep? > 1 h; 1 h; 50 min; 40 min; 30 min; 20 min; 10 min; 5 min; < 5 min.

4.   How many awakenings have you had during sleep? 0; 1; 2; 3; > 3.

5.   How long did it take you to go back to sleep after each awakening? < 1 min; 1-5 min; 6-10 min; 11-20 min; > 20 min. (Please specify to what awakening . first, second, etc. . belongs the duration of wakefulness).

6.   Are you refreshed after sleep? Totally; partly; not refreshed.

7.   Was your sleep in the first part of the night: deep; moderate; superficial?

8.   Was your sleep in the second part of the night: deep; moderate; superficial?

9.   Have you experienced dreams during your sleep? Yes; no.

10.   Do you remember some of your dreams? Yes; no.

11.   Have you been: (a) an observer; (b) an active participant in your dream content?

12.   Your dreams were predominantly: pleasant; neutral; unpleasant; frightening.

13.   Does your sleep mood after sleep become: better than in the evening; the same; worse than in the evening?

References

Ansseau, M., Kupfer, D.J., Reynolds III, C.F., 1985. Internight variability of REM latency in major depression: implications for the use of REM latency as a biological correlate. Biol. Psychiatry 20, 489-505.

Benca, R.M., Obermeyer, W.H., Thisted, R.A., Gillin, J.C., 1992. Sleep and psychiatric disorders: a meta-analysis. Arch. Gen. Psychiatry 49, 651-668.

Benson, K.L., Zarcone, V.P., 1993. Rapid eye movement sleep eye movements in schizophrenia and depression. Arch. Gen. Psychiatry 50, 474-482.

Buysse,  D.J., Reynolds, C.F., Mank, T.M., Berman, J.R., Kupfer, D.J., 1989. The Pittsburgh sleep quality index: a new instrument for psychiatric practice and research. Psychiatry Res. 28, 193-213.

Dietzel, M., Saletu, ..., Lesch, O.M., Sieghart, W., Sehjerve, M., 1986. Light treatment in depressive illness. Polysomno-graphic, psychometric and neuroendocrinological findings. Eur. Neurol. 25, 93-103.

Hamilton, M., 1960. A rating scale for depression. J. Neurol. Neurosurg. Psychiatry 23, 56-62.

Hemmeter, U., Seifritz, E., Hatzinger, M., Muller, M.J., Hols-boer-Trachsler, E., 1995. Serial partial sleep deprivation as adjuvant treatment of depressive insomnia. Prog. Neuro-Psychopharmacol. Biol. Psychiatry 19, 593-602.

Hong Ch, Ch.-h., Potkin, S.G., Antrobus, J.S., Dow, B.M., Callaghan, G.M., Gillin, J.Ch., 1997. REM sleep eye movement counts correlate with visual imagery in dreaming: a pilot study. Psychophysiology 14, 377-381.

Keklund, G., Akerstedt, ".., 1997. Objective components of individual differences in subjective sleep quality. J. Sleep Res. 6, 217-220.

Kramer, M., Roth, ".., 1973. A comparison of dream content in dream reports of schizophrenic and depressive patients groups. Compr. Psychiatry 14, 325-329.

Kupfer, D.J., Edwards, D.J., Spiker, D.G., Holzer, B.C., Coble, P., 1980. MAO activity and EEG sleep in primary depression. Psychiatry Res. 1, 241-247.

Kupfer,  D.J., Frank, E., Grochocinski, V.J., Gregor, M., McEacharon, A.B., 1988. Electroencephalographic sleep profiles in recurrent depression. Arch. Gen. Psychiatry 45, 678-681.

Lee, J.H., Reynolds C.F., Hoch C.C., Buysse D.J., Masumdar S., George C.J., Kupfer, D.J. 1993. Electroencephalographic sleep in recently remitted, elderly depressed patients in double-blind placebo-maintenance therapy. Neuropsychopharmacology, 8, 143-150.

Paiva, ".., Arriaga, F., Wanquier, A., Lara, E., Largo, R., Leitao, J.W., 1988. Effects of ritanserin on sleep disturbances of dysthymic patients. Psychopharmacology 96, 395-399.

Parrino, L., Ipaggiari, M.C., Boselli, M., Di-Giovanni, G., Terzano, M.G., 1994. Clinical and polysomnographic effects of trazodone CR in chronic insomnia associated with dys-thymia. Psychopharmacology 116, 389-395.

Pasternak, R.E., Reynolds, C.F., Hoch, C.C. et al., 1992. Sleep in spousally bereaved elders with subsyndromal depressive symptoms. Psychiatry Res. 43, 43-53.

Pasternak, R.E., Reynolds, C.F., Hoch, C.C. et al., 1994. Sleep in bereavement-related depression during and after phar-macotherapy with nortriptyline. J. Geriatr. Psychiatry Neurol. 7, 69-73.

Rechtschaffen, A., Kales, A. (Eds.), 1968. A Manual of Standardized Terminology, Techniques and Scoring System for Sleep Stages of Human Subjects. Washington, D.C. National Institute of Health, N.I.H. Public, p. 204.

Reynolds, C.F., Kupfer, D.J., 1987. Sleep research in affective illness: state of art circa 1987. Sleep 10, 199-215.

Rotenberg V.S. (1982) The adaptive function of sleep. Moscow, Nauka (in Russian).

Rotenberg, V.S., 1988. Functional deficiency of REM sleep and its role in the pathogenesis of neurotic and psychosomatic disturbances. Pavlov. J. Biol. Sci. 23, 1-3.

Rotenberg, V.S., 1993. The estimation of sleep quality in different stages and sleep cycles. J. Sleep Res. 2, 17-20.

Rotenberg, V.S., Hadjez, J., Kimhi, R. et al., 1997a. First night effect in depression: new data and a new approach. Biol. Psychiatry 42, 267-274.

Rotenberg, V.S., Kayumov, L., Indursky, P. et al., 1997b. REM sleep in depressed patients: different attempts to achieve adaptation. J. Psychosom. Res. 42, 565-575.

Rotenberg, V.S., Indursky P., Kimhi R., et al. 1999. The relationship between objective sleep variables and subjective sleep estimation in schizophrenia. Int. J. Psychiatry Clin. Pract. (in press).