43
Psychiatry Research, 43:43-53 Elsevier
Sleep in Spousally Bereaved Depressive Symptoms
Elders
With
Subsyndromal
Rona E. Pasternak, Charles F. Reynolds III, Carolyn C. Hoch, Daniel S. Buysse, Maryann Schlernitzauer, Maryann Machen, and David J. Kupfer Received January 16, 1992; revised version received April 7, 1992; accepted June 6, 1992. Abstract. Spousal bereavement in late life frequently leads to major depression. However, many people suffer from “minor” depressive symptoms that entail considerable suffering even in the absence of syndromal major depression. We describe longitudinal electroencephalographic (EEG) sleep and clinical evaluations in 14 elderly, recently spousally bereaved subjects who were experiencing subsyndromal depressive symptoms. While subjects did not meet diagnostic criteria for syndromal major depression, they did have mildly elevated scores on the Hamilton Rating Scale for Depression (mean = 10.6, range = 8-16) at the time of initial sleep studies (T,), which were carried out, on average, 5.5 months after loss of the spouse. Entry into the study was limited to volunteers who did not have a personal history of major depression or psychiatric disorder. Twelve subjects underwent followup clinical and EEG sleep evaluations (T2), 9.9 months after spousal loss. Fifty percent continued to show depressive symptoms at 6month followup. Test-retest comparisons of sleep and clinical measures were made with a group of sex- and age-matched control subjects who were neither bereaved nor depressed. EEG sleep measures did not significantly correlate with time from loss of spouse, severity of depressive symptoms, or subjective sleep quality. Analysis of variance with repeated measures detected a significant group X time interaction effect for delta sleep ratio (decreasing in controls but increasing in the bereaved). Key Words. bereavement.
Polysomnography,
electroencephalogram,
minor
depression,
Spousal loss through death is a common occurrence in older adults, estimated at’ 1.6% for males and 3.0% for women annually (Murrell at al., 1984). Approximately lo-20% of these people are known to suffer from unremitting depression during the first year of bereavement (Clayton et al., 1972; Bruce et al., 1990; Zisook and Schuchter, 1991) and even more may suffer from “minor” or subsyndromal depression. Very little is known about the clinical and physiological effects of subsyndromal depression associated with major life stressors, such as spousal bereavement.
Rona E. Pasternak, M.D., Carolyn C. Hoch, Ph.D., and Daniel S. Buysse, M.D., are Assistant Professors, and Charles F. Reynolds III, M.D., and David J. Kupfer, M.D., are Professors in the Department of Psychiatry, University of Pittsburgh. Maryann Schlernitzauer, B.S.N., is Research Project Coordinator, and Maryann Machen, B.S., is Research Associate in the Sleep and Chronobiology Center, Department of Psychiatry. (Reprint requests to Dr. C.F. Reynolds III, Western Psychiatric Institute and Clinic, 3811 O’Hara St., Pittsburgh, PA 15213-2593, USA.) 0165-1781/92/$05.00
@ 1992 Elsevier Scientific
Publishers
Ireland
Ltd.
44
Sleep disturbance is a clinical problem in spousal bereavement that can be persistent and debilitating (Clayton, 1980) and may predict the subsequent development of major depression (Ford and Kamerow, 1989). Sleep changes in bereavementrelated major depression include diminished sleep efficiency, shortened rapid eye movement (REM) latency, and lowered delta sleep ratio, while electroencephalographic (EEG) sleep in spousally bereaved elders without depression is similar to that of healthy nonbereaved, nondepressed controls (Reynolds et al., 1992). Furthermore, nortriptyline may be clinically useful in relieving depressive symptoms, including sleep disturbance in patients with bereavement-related depression (Pasternak et al., 1991). In a recent study that explored sleep changes in mid-life during a stressful life event, Cartwright and Wood (1991) discovered sleep changes consistent with depression in a subgroup of nondepressed subjects who were experiencing a divorce. Their findings suggest that the risk of developing depression is heightened, at least until the stressor is resolved. To our knowledge, however, no investigations of sleep changes specifically in subsyndromal depression have been reported, despite the theoretical and clinical interest of doing so. While the EEG sleep changes associated with major depression are well known and appear to be trait-like (i.e., persisting into remission), we do not know if such changes occur on a continuum related to severity, or are seen only in cases that cross a certain threshold of severity. Subsyndromal depression in the context of a major mood-disturbing event, like the loss of a spouse through death, provides a useful opportunity to address this question. Accordingly, the purpose of this report is to describe the clinical profile and EEG sleep correlates of subsyndromal depression in elderly bereaved subjects. Later reports will describe the clinical course and EEG sleep changes over time in an effort to further our understanding of how the stress of a major loss might affect biologic indicators, such as sleep, in these elders.
Methods Subjects. The current sample (n = 14) is part of a larger group of elderly volunteers who recently lost a spouse and are participating in an investigation of electroencephalographic (EEG) sleep changes associated with spousal bereavement (Reynolds et al., 1992). Twelve have undergone both initial (T,) and followup (T,) assessments. Subjects were recruited through media announcements (n = 12), word of mouth (n = 1), and oral presentation (n = 1). They were required to be 60 to 80 years of age, to have no unstable medical disorder or take any medications that would confound sleep assessment, to have no prior history of psychiatric disorder (including alcohol abuse), and no evidence of dementia. As well, each must have suffered from conjugal bereavement within the last year. An extensive baseline evaluation included the following measures: (1) a structured psychiatric interview, the Lifetime version of the Schedule for Affective Disorders and Schizophrenia (SADS-L; Spitzer et al., 1978), to determine the presence or absence of current and past major depression; (2) baseline health status determination on the basis of complete medical history, physical examination, and routine laboratory tests (including complete blood count with differential, Astra 7, amylase, serum glutamic-oxaloacetic transaminase, serum glutamicpyruvic transaminase, bilirubin [total and direct], cholesterol, calcium, uric acid, thyroid profile, folate, vitamin B,,, rapid plasma reagin, urinalysis, chest x-ray, and electrocardiogram); and (3) a battery of rater- and self-administered scales to determine the
45 severity of depressive symptoms (17-item Hamilton Rating Scale for Depression [HRSD; Hamilton, 19601 and Beck Depression Inventory [BDI; Beck et al., 1961]), functional impairment (Global Assessment Scale [GAS; Endicott et al., 19761, other symptoms of psychopathological distress (Brief Symptom Inventory [BSI; Derogatis and Melisaratos, 19831) intensity of grief (Texas Revised Inventory of Grief [TRIG; Fashingbauer, 1981]), grief intensity focusing on feelings of numbness and separation anxiety (Jacobs et al., 1986), subjective sleep quality (Pittsburgh Sleep Quality Index [PSQI; Buysse et al., 1989]), medical burden (Cumulative Illness Rating Scale [CIRS; Linn et al., 1968; Miller et al., 1992]), stability of daily social rhythms (Social Rhythm Metric [SRM; Monk et al., 19901) perceived social support (Interpersonal Support Evaluation List [ISEL; Williamson and Schulz, 19901) and evaluation of cognitive status (Folstein Mini-Mental State Examination [Folstein et al., 19751). To qualify for a diagnosis of subsyndromal depression, subjects had no present or past major depression on the basis of the Research Diagnostic Criteria (Spitzer et al., 1978) and had stable (for at least 2 weeks) HRSD scores in the range of 8-16. Thus, these volunteers were suffering from depressive symptoms of a lesser severity than that seen in bereavement-related major depression, but to a greater extent than seen in bereavement without depression (HRSD scores < 7). Healthy control subjects were recruited and underwent extensive medical and psychiatric evaluations similar to those that are described above. Elderly control subjects were required to be in good physical health with no sleep complaints and to have no personal or family history of psychiatric disorder in a first degree relative. In addition, control subjects were required to have a score < 7 on the HRSD and a score > 28 on the Folstein Mini-Mental State Examination. To clarify the degree of medical burden, total medication use and high CIRS scores in any one category (i.e., 3 or 4) were documented. At T,, three control subjects and seven bereaved subjects used prescription medication. At T,, three control subjects and five bereaved subjects used medication. Two control subjects and four bereaved subjects had one or more CIRS category scores of 3 or 4. Assessment of EEG Sleep. Subjective sleep quality (PSQI) and laboratory sleep assessments were performed at two points in each group. Bereaved subjects underwent T, studies at 5.5 (SD = 2.9) months and T, studies at 9.9 (SD = 1.5) months from the loss of the spouse. The interval from T, to T, in control subjects was 12.4 (SD = 0.4) months. Two weeks before sleep studies, all subjects were instructed to use no alcohol or psychotropic drugs and to keep a diary in which they were to record sleep-wake schedule and medication use. Subjects reported alcohol intake as follows: At T,, 4/ 14 control subjects consumed alcohol, with a maximum of three drinks, and 2/ 14 bereaved subjects consumed alcohol, with a maximum of one drink. At T,, 2/ 12 control subjects reported alcohol use, with a maximum of two drinks, and 4/ 12 bereaved subjects reported alcohol use, with a maximum of three drinks. Four control subjects and no bereaved subject had bed partners. Previously described methods (Reynolds et al., 1992) were used to carry out 3 consecutive nights of laboratory-based polysomnographic studies after the 2-week psychotropic-drug and alcohol-free intervals. Subjects were studied at their habitual sleep times and were also monitored for sleep apnea and periodic limb movements on night 1. Records were scored manually by polysomnographic technologists who achieved at least 85% agreement among the scores and who were unaware of the diagnostic status of the subjects. Automated (period) analysis of delta wave activity in the EEG was performed according to previously published procedures (Kupfer et al., 1984). Averaged data from nights 2 and 3 were used in the analysis. Statistical Analysis. Group t tests were used to compare clinical and sleep variables between the two groups. The K statistic was used to examine further the distributions of PSQI scores and delta sleep ratio values between control subjects and bereaved subjects. A cutoff score of 5 was chosen on the basis of previous findings with the PSQI (Buysse et al., 1989) and a cutoff score of 1.3 for the delta sleep ratio was selected on the basis of the distribution of
46 Spearman rank order correlation coefficients showed that sleep quality, sleep efficiency, REM latency, and delta sleep ratio were related to (1) HRSD ratings, (2) sleep
scores.
quality scores (PSQI), and (3) time since loss of spouse. Repeated measures analyses of variance (ANOVA) were used to examine group and time effects and group X time interactions for the distribution of clinical and sleep measures. Longitudinal Course. Subjects were contacted on a monthly The rater- and self-administered scales described above were of depressive symptoms over time. Volunteers were reclassified as subsyndromal or remitted on the basis of HRSD scores. If (at least two consecutive ratings) < 6, patients were classified
basis for followup evaluations. used to track the clinical course at 6 months after sleep studies HRSD scores were consistently as remitted.
Results of Demographic and Clinical Measures (Table 1). At T,, bereaved subjects demonstrated intense grieving (TRIG score = 50.4, SD = 6.3). Clinical differences between the bereaved group and the control group included mild depressive symptoms in the former, as evidenced by HRSD (p < 0.0001) and BDI scores (p < 0.0001); functional impairment (GAS: p < 0.0001); higher self-reported ratings of psychopathological symptoms (BSI): (p < 0.01); and diminished social support (ISEL: p < 0.05).Cumulative medical burden (CIRS) and stability of social rhythms (SRM) did not differ between bereaved subjects and control subjects. Followup data, evaluated by repeated measures ANOVA, detected a significant group X time effect for HRSD scores (p < O.Ol), reflecting some clinical improvement over time in the experimental group, but stability in the control subjects. Comparison
of Sleep Quality and EEG Sleep Measures (Table 2). As reflected by increased PSQI scores, subjective sleep quality was worse in the bereaved group (p < 0.01) at T,. Fig. 1 presents the distribution of PSQI scores, along with the cutoff score of 5. All of the subjects in the subsyndromal group, but only 4 of 14 control subjects, had PSQI scores Z 5 (K = 0.71; p = 0.0001). Bereaved subjects differed from control subjects in PSQI factors that measured global sleep quality, prolonged sleep latency, wakefulness after sleep onset, and symptoms of daytime dysfunction such as fatigue. Severity of sleep complaints (PSQI), severity of depressive symptoms (HRSD), and time since loss were not significantly correlated with EEG sleep measurements (sleep efficiency, REM latency, and delta sleep ratio) in the subsyndromally depressed elderly subjects. A trendworthy correlation was found for HRSD scores and sleep efficiency (p = -0.52, p < 0.06). There was no correlation in the control subjects between PSQI or HRSD scores with sleep measures (including sleep efficiency, REM latency, and delta sleep ratio). The two groups did not differ on any visually scored sleep measures. Automated delta sleep ratio tended to be lower in the bereaved group (p < 0.10).(Delta sleep ratio is the ratio of delta wave [0.5-3Hz, 75-200 PV] counts per minute [cpm] in the first non-REM period to delta wave cpm in the second non-REM period [Kupfer et al., 19901.) The distribution of delta sleep ratios in each group, on the basis of a post hoc cutoff score > 1.3 (Fig. 2), showed that 9 of 14 control subjects, but only 3 of 14 subjects in the subsyndromal group, had delta sleep ratios 3 1.3 (K = 0.43, p = 0.01I). Comparison
47 The repeated measures ANOVA of followup sleep measures detected a significant group X time interaction for delta sleep ratio @ < 0.05) that reflected an increase among the subsyndromally depressed subjects and a decrease in control subjects. Stage 2% also behaved differently in the two groups (p < 0.01) (i.e., increasing in the control subjects and decreasing in the bereaved subjects). Longitudinal Course. Ten subjects have completed the 6-month followup evaluation. At 6 months, five subjects have experienced remission of their depressive symptoms while five continue to show subsyndromal symptoms of depression. All control subjects have remained well without significant psychiatric symptoms.
Discussion Spousally bereaved elderly patients with subsyndromal depressive symptoms showed clinically significant impairment in sleep quality, though not to the extent seen in bereaved subjects with major depression (Reynolds et al., 1992). In general, physiological changes during sleep that are reliably seen in major syndromal depression (diminished REM sleep latency, prolonged first REM sleep period, and impaired sleep efficiency) were absent, except for the trend for a decreased delta sleep ratio and the apparent improvement of delta sleep ratio over time in the subsyndromally depressed group. This change cannot be attributed to differences in medication and alcohol intake, since these variables were similar in the two groups. An attenuated delta sleep ratio signifies a low rate of delta-wave generation in the first non-REM period, which has been found in nonbereavement-related major depression in both middle (Kupfer et al., 1990) and late life (Reynolds et al., 1991). Furthermore, in late-life spousal bereavement with syndromal major depression, the severity of depression has been shown to correlate with diminished delta sleep ratio (Reynolds et al., 1992). Although visually scored measures of sleep show changes in relation to major depression, these changes do not exist along a continuum of severity in relation to subsyndromal depression. By contrast, the automated measure of slow wave activity suggests that there may be change in distribution of slow wave activity similar to that seen in major depression (whether bereavement-related or recurrent). This distribution may improve over time in subsyndromal depression. In a recent study of adjustment disorders of sleep among middle-aged women (mean age = 35.0) and men (mean age = 37.7) undergoing divorce, Cartwright and Wood (1991) reported that delta sleep increased at followup for those whose divorces were completed. REM latency was reduced and REM percent was elevated only in the depressed women; however, among the nondepressed women, those whose divorces remained incomplete at followup had lower delta, higher REM percent, and shorter REM latency than did those whose divorces were finalized. Previously, we have also noted increased REM percent, decreased REM latency, and decreased delta sleep ratio among elderly bereaved subjects with full syndromal (first-episode) depression, similar to the findings of Cartwright and Wood (1991). Our followup data in the subsyndromally depressed bereaved subjects also suggest improvement in the delta sleep ratio.
Time 1 (n= 14)
Present
-
-
Scale
-
Numbness
Separation Anxiety Center for Epidemiologic Studies Depression
Grief intensity
-
Past
-
-
-
-
-
Time 2 (n= 12)
3.7
Count
Texas Revised Inventory of Grief
1.9
4.6
Total 3.2
78
6:42
Habitual risetime
Cumulative Illness Rating Scale
40
2253
Habitual bedtime
4.5
5.1 19.6
2.2
2.5 .13.0
7.0 6.3
21.1 50.4
13.9
9.9
2.3
44.8
19.4
4.6
5.6
2.3
9.6
4.4
Time 2 (n= 12)
2.4
Time 1 (n= 14)
3.8
38
56
1.5
2.9
4.4
3111
2.9
7.0
SD
6.1
7:23
23:31
4.4
0.4
5.5
interval (mo)
Followup
13.8
Months from loss
Gender (male/female) 12.4
2.0
68.1
6.9
(yr)
14.0
Mean
Subsyndromal bereaved (n= 14) SD
68.2 3111
Control (n= 14)
Education
Mean
and clinical measures
Age (yr)
Table 1. Demographic
-3.73
-2.09
-0.80
-1.80
-0.53
t
Paired
-0.86
-1.14
-1.75
-1.76
0.23
0.05
t
0.003
NS
NS
NS
NS
P
NS
NS
NS
NS
NS
NS
R
’
Scale
Sleep Quality Index
Inventory
inventory
11.4
Belonging
l.p
Psychiatry Research, 43:43-53 Elsevier
Sleep in Spousally Bereaved Depressive Symptoms
Elders
With
Subsyndromal
Rona E. Pasternak, Charles F. Reynolds III, Carolyn C. Hoch, Daniel S. Buysse, Maryann Schlernitzauer, Maryann Machen, and David J. Kupfer Received January 16, 1992; revised version received April 7, 1992; accepted June 6, 1992. Abstract. Spousal bereavement in late life frequently leads to major depression. However, many people suffer from “minor” depressive symptoms that entail considerable suffering even in the absence of syndromal major depression. We describe longitudinal electroencephalographic (EEG) sleep and clinical evaluations in 14 elderly, recently spousally bereaved subjects who were experiencing subsyndromal depressive symptoms. While subjects did not meet diagnostic criteria for syndromal major depression, they did have mildly elevated scores on the Hamilton Rating Scale for Depression (mean = 10.6, range = 8-16) at the time of initial sleep studies (T,), which were carried out, on average, 5.5 months after loss of the spouse. Entry into the study was limited to volunteers who did not have a personal history of major depression or psychiatric disorder. Twelve subjects underwent followup clinical and EEG sleep evaluations (T2), 9.9 months after spousal loss. Fifty percent continued to show depressive symptoms at 6month followup. Test-retest comparisons of sleep and clinical measures were made with a group of sex- and age-matched control subjects who were neither bereaved nor depressed. EEG sleep measures did not significantly correlate with time from loss of spouse, severity of depressive symptoms, or subjective sleep quality. Analysis of variance with repeated measures detected a significant group X time interaction effect for delta sleep ratio (decreasing in controls but increasing in the bereaved). Key Words. bereavement.
Polysomnography,
electroencephalogram,
minor
depression,
Spousal loss through death is a common occurrence in older adults, estimated at’ 1.6% for males and 3.0% for women annually (Murrell at al., 1984). Approximately lo-20% of these people are known to suffer from unremitting depression during the first year of bereavement (Clayton et al., 1972; Bruce et al., 1990; Zisook and Schuchter, 1991) and even more may suffer from “minor” or subsyndromal depression. Very little is known about the clinical and physiological effects of subsyndromal depression associated with major life stressors, such as spousal bereavement.
Rona E. Pasternak, M.D., Carolyn C. Hoch, Ph.D., and Daniel S. Buysse, M.D., are Assistant Professors, and Charles F. Reynolds III, M.D., and David J. Kupfer, M.D., are Professors in the Department of Psychiatry, University of Pittsburgh. Maryann Schlernitzauer, B.S.N., is Research Project Coordinator, and Maryann Machen, B.S., is Research Associate in the Sleep and Chronobiology Center, Department of Psychiatry. (Reprint requests to Dr. C.F. Reynolds III, Western Psychiatric Institute and Clinic, 3811 O’Hara St., Pittsburgh, PA 15213-2593, USA.) 0165-1781/92/$05.00
@ 1992 Elsevier Scientific
Publishers
Ireland
Ltd.
44
Sleep disturbance is a clinical problem in spousal bereavement that can be persistent and debilitating (Clayton, 1980) and may predict the subsequent development of major depression (Ford and Kamerow, 1989). Sleep changes in bereavementrelated major depression include diminished sleep efficiency, shortened rapid eye movement (REM) latency, and lowered delta sleep ratio, while electroencephalographic (EEG) sleep in spousally bereaved elders without depression is similar to that of healthy nonbereaved, nondepressed controls (Reynolds et al., 1992). Furthermore, nortriptyline may be clinically useful in relieving depressive symptoms, including sleep disturbance in patients with bereavement-related depression (Pasternak et al., 1991). In a recent study that explored sleep changes in mid-life during a stressful life event, Cartwright and Wood (1991) discovered sleep changes consistent with depression in a subgroup of nondepressed subjects who were experiencing a divorce. Their findings suggest that the risk of developing depression is heightened, at least until the stressor is resolved. To our knowledge, however, no investigations of sleep changes specifically in subsyndromal depression have been reported, despite the theoretical and clinical interest of doing so. While the EEG sleep changes associated with major depression are well known and appear to be trait-like (i.e., persisting into remission), we do not know if such changes occur on a continuum related to severity, or are seen only in cases that cross a certain threshold of severity. Subsyndromal depression in the context of a major mood-disturbing event, like the loss of a spouse through death, provides a useful opportunity to address this question. Accordingly, the purpose of this report is to describe the clinical profile and EEG sleep correlates of subsyndromal depression in elderly bereaved subjects. Later reports will describe the clinical course and EEG sleep changes over time in an effort to further our understanding of how the stress of a major loss might affect biologic indicators, such as sleep, in these elders.
Methods Subjects. The current sample (n = 14) is part of a larger group of elderly volunteers who recently lost a spouse and are participating in an investigation of electroencephalographic (EEG) sleep changes associated with spousal bereavement (Reynolds et al., 1992). Twelve have undergone both initial (T,) and followup (T,) assessments. Subjects were recruited through media announcements (n = 12), word of mouth (n = 1), and oral presentation (n = 1). They were required to be 60 to 80 years of age, to have no unstable medical disorder or take any medications that would confound sleep assessment, to have no prior history of psychiatric disorder (including alcohol abuse), and no evidence of dementia. As well, each must have suffered from conjugal bereavement within the last year. An extensive baseline evaluation included the following measures: (1) a structured psychiatric interview, the Lifetime version of the Schedule for Affective Disorders and Schizophrenia (SADS-L; Spitzer et al., 1978), to determine the presence or absence of current and past major depression; (2) baseline health status determination on the basis of complete medical history, physical examination, and routine laboratory tests (including complete blood count with differential, Astra 7, amylase, serum glutamic-oxaloacetic transaminase, serum glutamicpyruvic transaminase, bilirubin [total and direct], cholesterol, calcium, uric acid, thyroid profile, folate, vitamin B,,, rapid plasma reagin, urinalysis, chest x-ray, and electrocardiogram); and (3) a battery of rater- and self-administered scales to determine the
45 severity of depressive symptoms (17-item Hamilton Rating Scale for Depression [HRSD; Hamilton, 19601 and Beck Depression Inventory [BDI; Beck et al., 1961]), functional impairment (Global Assessment Scale [GAS; Endicott et al., 19761, other symptoms of psychopathological distress (Brief Symptom Inventory [BSI; Derogatis and Melisaratos, 19831) intensity of grief (Texas Revised Inventory of Grief [TRIG; Fashingbauer, 1981]), grief intensity focusing on feelings of numbness and separation anxiety (Jacobs et al., 1986), subjective sleep quality (Pittsburgh Sleep Quality Index [PSQI; Buysse et al., 1989]), medical burden (Cumulative Illness Rating Scale [CIRS; Linn et al., 1968; Miller et al., 1992]), stability of daily social rhythms (Social Rhythm Metric [SRM; Monk et al., 19901) perceived social support (Interpersonal Support Evaluation List [ISEL; Williamson and Schulz, 19901) and evaluation of cognitive status (Folstein Mini-Mental State Examination [Folstein et al., 19751). To qualify for a diagnosis of subsyndromal depression, subjects had no present or past major depression on the basis of the Research Diagnostic Criteria (Spitzer et al., 1978) and had stable (for at least 2 weeks) HRSD scores in the range of 8-16. Thus, these volunteers were suffering from depressive symptoms of a lesser severity than that seen in bereavement-related major depression, but to a greater extent than seen in bereavement without depression (HRSD scores < 7). Healthy control subjects were recruited and underwent extensive medical and psychiatric evaluations similar to those that are described above. Elderly control subjects were required to be in good physical health with no sleep complaints and to have no personal or family history of psychiatric disorder in a first degree relative. In addition, control subjects were required to have a score < 7 on the HRSD and a score > 28 on the Folstein Mini-Mental State Examination. To clarify the degree of medical burden, total medication use and high CIRS scores in any one category (i.e., 3 or 4) were documented. At T,, three control subjects and seven bereaved subjects used prescription medication. At T,, three control subjects and five bereaved subjects used medication. Two control subjects and four bereaved subjects had one or more CIRS category scores of 3 or 4. Assessment of EEG Sleep. Subjective sleep quality (PSQI) and laboratory sleep assessments were performed at two points in each group. Bereaved subjects underwent T, studies at 5.5 (SD = 2.9) months and T, studies at 9.9 (SD = 1.5) months from the loss of the spouse. The interval from T, to T, in control subjects was 12.4 (SD = 0.4) months. Two weeks before sleep studies, all subjects were instructed to use no alcohol or psychotropic drugs and to keep a diary in which they were to record sleep-wake schedule and medication use. Subjects reported alcohol intake as follows: At T,, 4/ 14 control subjects consumed alcohol, with a maximum of three drinks, and 2/ 14 bereaved subjects consumed alcohol, with a maximum of one drink. At T,, 2/ 12 control subjects reported alcohol use, with a maximum of two drinks, and 4/ 12 bereaved subjects reported alcohol use, with a maximum of three drinks. Four control subjects and no bereaved subject had bed partners. Previously described methods (Reynolds et al., 1992) were used to carry out 3 consecutive nights of laboratory-based polysomnographic studies after the 2-week psychotropic-drug and alcohol-free intervals. Subjects were studied at their habitual sleep times and were also monitored for sleep apnea and periodic limb movements on night 1. Records were scored manually by polysomnographic technologists who achieved at least 85% agreement among the scores and who were unaware of the diagnostic status of the subjects. Automated (period) analysis of delta wave activity in the EEG was performed according to previously published procedures (Kupfer et al., 1984). Averaged data from nights 2 and 3 were used in the analysis. Statistical Analysis. Group t tests were used to compare clinical and sleep variables between the two groups. The K statistic was used to examine further the distributions of PSQI scores and delta sleep ratio values between control subjects and bereaved subjects. A cutoff score of 5 was chosen on the basis of previous findings with the PSQI (Buysse et al., 1989) and a cutoff score of 1.3 for the delta sleep ratio was selected on the basis of the distribution of
46 Spearman rank order correlation coefficients showed that sleep quality, sleep efficiency, REM latency, and delta sleep ratio were related to (1) HRSD ratings, (2) sleep
scores.
quality scores (PSQI), and (3) time since loss of spouse. Repeated measures analyses of variance (ANOVA) were used to examine group and time effects and group X time interactions for the distribution of clinical and sleep measures. Longitudinal Course. Subjects were contacted on a monthly The rater- and self-administered scales described above were of depressive symptoms over time. Volunteers were reclassified as subsyndromal or remitted on the basis of HRSD scores. If (at least two consecutive ratings) < 6, patients were classified
basis for followup evaluations. used to track the clinical course at 6 months after sleep studies HRSD scores were consistently as remitted.
Results of Demographic and Clinical Measures (Table 1). At T,, bereaved subjects demonstrated intense grieving (TRIG score = 50.4, SD = 6.3). Clinical differences between the bereaved group and the control group included mild depressive symptoms in the former, as evidenced by HRSD (p < 0.0001) and BDI scores (p < 0.0001); functional impairment (GAS: p < 0.0001); higher self-reported ratings of psychopathological symptoms (BSI): (p < 0.01); and diminished social support (ISEL: p < 0.05).Cumulative medical burden (CIRS) and stability of social rhythms (SRM) did not differ between bereaved subjects and control subjects. Followup data, evaluated by repeated measures ANOVA, detected a significant group X time effect for HRSD scores (p < O.Ol), reflecting some clinical improvement over time in the experimental group, but stability in the control subjects. Comparison
of Sleep Quality and EEG Sleep Measures (Table 2). As reflected by increased PSQI scores, subjective sleep quality was worse in the bereaved group (p < 0.01) at T,. Fig. 1 presents the distribution of PSQI scores, along with the cutoff score of 5. All of the subjects in the subsyndromal group, but only 4 of 14 control subjects, had PSQI scores Z 5 (K = 0.71; p = 0.0001). Bereaved subjects differed from control subjects in PSQI factors that measured global sleep quality, prolonged sleep latency, wakefulness after sleep onset, and symptoms of daytime dysfunction such as fatigue. Severity of sleep complaints (PSQI), severity of depressive symptoms (HRSD), and time since loss were not significantly correlated with EEG sleep measurements (sleep efficiency, REM latency, and delta sleep ratio) in the subsyndromally depressed elderly subjects. A trendworthy correlation was found for HRSD scores and sleep efficiency (p = -0.52, p < 0.06). There was no correlation in the control subjects between PSQI or HRSD scores with sleep measures (including sleep efficiency, REM latency, and delta sleep ratio). The two groups did not differ on any visually scored sleep measures. Automated delta sleep ratio tended to be lower in the bereaved group (p < 0.10).(Delta sleep ratio is the ratio of delta wave [0.5-3Hz, 75-200 PV] counts per minute [cpm] in the first non-REM period to delta wave cpm in the second non-REM period [Kupfer et al., 19901.) The distribution of delta sleep ratios in each group, on the basis of a post hoc cutoff score > 1.3 (Fig. 2), showed that 9 of 14 control subjects, but only 3 of 14 subjects in the subsyndromal group, had delta sleep ratios 3 1.3 (K = 0.43, p = 0.01I). Comparison
47 The repeated measures ANOVA of followup sleep measures detected a significant group X time interaction for delta sleep ratio @ < 0.05) that reflected an increase among the subsyndromally depressed subjects and a decrease in control subjects. Stage 2% also behaved differently in the two groups (p < 0.01) (i.e., increasing in the control subjects and decreasing in the bereaved subjects). Longitudinal Course. Ten subjects have completed the 6-month followup evaluation. At 6 months, five subjects have experienced remission of their depressive symptoms while five continue to show subsyndromal symptoms of depression. All control subjects have remained well without significant psychiatric symptoms.
Discussion Spousally bereaved elderly patients with subsyndromal depressive symptoms showed clinically significant impairment in sleep quality, though not to the extent seen in bereaved subjects with major depression (Reynolds et al., 1992). In general, physiological changes during sleep that are reliably seen in major syndromal depression (diminished REM sleep latency, prolonged first REM sleep period, and impaired sleep efficiency) were absent, except for the trend for a decreased delta sleep ratio and the apparent improvement of delta sleep ratio over time in the subsyndromally depressed group. This change cannot be attributed to differences in medication and alcohol intake, since these variables were similar in the two groups. An attenuated delta sleep ratio signifies a low rate of delta-wave generation in the first non-REM period, which has been found in nonbereavement-related major depression in both middle (Kupfer et al., 1990) and late life (Reynolds et al., 1991). Furthermore, in late-life spousal bereavement with syndromal major depression, the severity of depression has been shown to correlate with diminished delta sleep ratio (Reynolds et al., 1992). Although visually scored measures of sleep show changes in relation to major depression, these changes do not exist along a continuum of severity in relation to subsyndromal depression. By contrast, the automated measure of slow wave activity suggests that there may be change in distribution of slow wave activity similar to that seen in major depression (whether bereavement-related or recurrent). This distribution may improve over time in subsyndromal depression. In a recent study of adjustment disorders of sleep among middle-aged women (mean age = 35.0) and men (mean age = 37.7) undergoing divorce, Cartwright and Wood (1991) reported that delta sleep increased at followup for those whose divorces were completed. REM latency was reduced and REM percent was elevated only in the depressed women; however, among the nondepressed women, those whose divorces remained incomplete at followup had lower delta, higher REM percent, and shorter REM latency than did those whose divorces were finalized. Previously, we have also noted increased REM percent, decreased REM latency, and decreased delta sleep ratio among elderly bereaved subjects with full syndromal (first-episode) depression, similar to the findings of Cartwright and Wood (1991). Our followup data in the subsyndromally depressed bereaved subjects also suggest improvement in the delta sleep ratio.
Time 1 (n= 14)
Present
-
-
Scale
-
Numbness
Separation Anxiety Center for Epidemiologic Studies Depression
Grief intensity
-
Past
-
-
-
-
-
Time 2 (n= 12)
3.7
Count
Texas Revised Inventory of Grief
1.9
4.6
Total 3.2
78
6:42
Habitual risetime
Cumulative Illness Rating Scale
40
2253
Habitual bedtime
4.5
5.1 19.6
2.2
2.5 .13.0
7.0 6.3
21.1 50.4
13.9
9.9
2.3
44.8
19.4
4.6
5.6
2.3
9.6
4.4
Time 2 (n= 12)
2.4
Time 1 (n= 14)
3.8
38
56
1.5
2.9
4.4
3111
2.9
7.0
SD
6.1
7:23
23:31
4.4
0.4
5.5
interval (mo)
Followup
13.8
Months from loss
Gender (male/female) 12.4
2.0
68.1
6.9
(yr)
14.0
Mean
Subsyndromal bereaved (n= 14) SD
68.2 3111
Control (n= 14)
Education
Mean
and clinical measures
Age (yr)
Table 1. Demographic
-3.73
-2.09
-0.80
-1.80
-0.53
t
Paired
-0.86
-1.14
-1.75
-1.76
0.23
0.05
t
0.003
NS
NS
NS
NS
P
NS
NS
NS
NS
NS
NS
R
’
Scale
Sleep Quality Index
Inventory
inventory
11.4
Belonging
l.p
