Abstract
Depression can be conceptualised according to two different paradigms: specific nosological entities as those defined in the Diagnosis and Statistical Manual of Mental Disorders, fourth edition (DSM-IV) and the general phenomenon of depressive symptoms [1]. Although most basic research and clinical investigations rely on the former paradigm, the utilisation of the latter has generated important findings particularly in epidemiological studies. In fact, research has shown that individuals with depressive symptoms, who otherwise do not fulfil the DSM-IV criteria for depressive disorders, exhibit more impairment in social functions relative to persons without depressive symptoms [2].
In a study investigating the functioning and well-being of depressed patients, Wells and colleagues [1] reported that individuals with depressive symptoms were characterised by lower physical, social, and role functioning, poorer perceived current health, and more bodily pain than healthy individuals.
Furthermore, Judd and colleagues [2] suggested that the presence of even a single significant depressive symptom is associated with increased utilisation of medical, psychiatric, and emergency services, episodes of attempted suicide, and receiving disability benefits. The 1-month point prevalence of depressive symptoms in the three-site Epidemiologic Catchment Area study (EcA, Baltimore, Durham and Los Angeles) community sample is 4.4%, while the prevalence of major depression is 2.3% [2]. Depressive symptoms tend to persist beyond 1 year but 10% of the cases proceed toward major depression [3]. Significant prevalence, serious morbidity, and a tendency to progress to major depression suggest that depressive symptoms pose a significant challenge to public health.
Depressed mood, regarded as an essential feature in the diagnosis of major depression, is present in 94% of the episodes [4]. The role of depressed mood in subthreshold depressions is less certain. Judd and colleagues [4] suggest that patients showing depressive symptoms without depressed mood should be included under the category of subsyndromal symptomatic depression. However, Broadhead and colleagues [3] showed that depressive symptoms without accompanying mood disturbance were not significantly related to disability and were so common as to be difficult to classify as a mental disorder.
Although most depressive symptoms are reported subjectively, sleep disturbances and psychomotor retardation can be objectively measured in clinical settings. Sleep disturbances have been objectively evaluated since the introduction of the polysomnography [5]. For psychomotor retardation, the turning point was the development of the Depressive Retardation Rating Scale [6], the first instrument ever designed to assess retardation independently of any other affective symptoms which, however, still relies on the patient's self-report. The CORE system, an 18-item assessment schedule for observed (objective) psychomotor signs, was later developed by Parker and colleagues [7–8]. Both systems were shown to predict ECT response [9].
In recent years, wrist actigraphy has been increasingly utilised as an alternative to other methods traditionally used to assess sleep and activity in samples of clinical patients and research subjects. There is ample evidence suggesting that wrist actigraphy can be used to study sleep patterns and psychomotor disturbances in depression [10],[11]. Wrist actigraphs are relatively inexpensive and well tolerated [12],[13]. Actigraphs are fairly non-intrusive and can provide a large amount of clinical data, a potential not fully explored to date.
We hypothesised that actigraphy could be used to monitor sleep and motor activity abnormalities that often accompany mood disturbances. To test this hypothesis, we examined whether depressed mood, as coded in the Alzheimer's Disease Assessment Scale, would be associated with actigraph-monitored daytime activity level and sleep/wake parameters in a non-psychiatric sample.
Method
Volunteers were recruited through a mailed prospectus and advertisement regarding a cross-sectional actigraphic study conducted at the Psychophysiology Laboratory of the City University of New York (CUNY). Responders were phoned to determine if they would participate after the study protocol was described to them. Participants comprised 32 individuals ranging from 18 to 79 years of age (mean age = 50.00, SD= 23.97, men = 41%, women = 59%), who reported no medical or psychiatric problem on initial interview. Volunteers exhibited normal cognitive ability as assessed by the Mini Mental State Exam (MMSE, mean = 27.03, SD = 3.86). Volunteers resided at home and provided written informed consent under the supervision of the Ethics Committee at the College of Staten Island, CUNY.
Procedures
Consenting volunteers received a home visit for delivery of the forms and recording instrument used in the study. They were monitored for 5 days throughout which they provided daily information regarding sleepiness and general mood states using the Stanford Sleepiness Scale and the Visual Analog Scale. Sleep and wakefulness data were acquired continuously with an actigraph (Gaehwiler Electronics, Hombrechtikon, Switzerland, see instrumentation), worn for 5 days, while daily activities were recorded in a sleep log. On the last day of the recording, volunteers received a neuropsychological evaluation including the Alzheimer's Disease Assessment Scale [14].
The Mini Mental State Exam was used as an indicator of global cognitive ability. One of the objectives of the study was to determine whether individuals across age groups were comparable cognitively and had similar mood profile. We used the Alzheimer's Disease Assessment Scale because it provided the assessment of both of these parameters. The overall score (mean = 11.46, SD = 15.79) from the Alzheimer's Disease Assessment Scale confirmed the findings from the MMSE, indicating that participants did not exhibit debilitating cognitive impairment. Only two individuals expressed symptoms of Alzheimer's disease. When these two cases were excluded from the analysis, the results were the same.
Furthermore, the Alzheimer's Disease Assessment Scale provided a clear objective method for assessing depressed mood, which in our study was defined as an expression of sadness, discouragement or feeling down that was further queried by the examiner. We favour the point of view that, after controlling for comorbid physical illness and actual cognitive impairment, depression in late life is not symptomatically different from depression at younger ages [15]. Therefore, we feel that the use of the ‘depressed mood’ item, as a measure of depression across this age range, did not represent a major methodological limitation.
Instrumentation and statistical analyses
Wrist actigraphy was used for objective recording of habitual activity and sleep/wake patterns [12],[13]. Actigraphic data were analysed with the Actigraph Data Analysis Software, which provides reliable estimates of sleep and wakefulness in normal persons and in individuals with insomnia [16–18]. Other investigators have also shown that actigraphy is useful in discriminating sleep from wakefulness [19],[20].
Actigraphic data were first visually inspected to ascertain actual time in bed. Time in bed was synchronised or adjusted by a trained scorer in cases where there was a discrepancy between reported bedtime and observed actigraphic counts. Data were then analysed to yield a number of sleep/wake parameters. Derived variables included daytime activity level, total sleep time, sleep onset latency, transition from sleep to wakefulness, wake after sleep onset, and time in bed, which were averaged over the 5-day recording period before final computations were made.
The primary objective of our statistical analysis was to determine which actigraphic parameters were associated with depressed mood by employing a multiple logistic regression model. Depressed mood, one of the non-cognitive items of the Alzheimer's Disease Assessment Scale, was assessed by a trained examiner. This assessment involved the examiner inquiring whether the volunteer had been sad, discouraged, and down. If a positive response was obtained, further inquiry into the severity and pervasiveness of the mood, loss of interest or pleasure in activities, and reactivity to environmental events was made. The volunteer was further assessed for depressed faces and the ability to respond to encouragement and jokes. Each volunteer received a score from 1, indicating slight dysphoria which was clinically significant through 5, representing a pervasive and severe degree of dysphoric mood; total lack of reactivity; and pervasive loss of interest or pleasure [14]. To build the logistic regression model, depressed mood was recoded with 0, indicating the absence of depressed mood and 1, representing the presence of depressed mood.
Methods
Means and standard deviations for actigraphic parameters that characterise the sample are presented in Table 1. Analysis revealed that 25% of the participants exhibited depressed mood. Significant associations were noted between age and depressed mood, total time in bed, total sleep time, and wakefulness after sleep onset (r = 0.62, p < 0.001; r = 0.47, p < 0.01; r = 0.43, p < 0.05; and r = 0.41, p < 0.05, respectively). However, in this data set age was not significantly associated with daytime activity level or circadian activity amplitude (r = 0.16, NS; and r = −0.16, NS, respectively).
Actigraphic sleep/wake parameters in a non-psychiatric sample
The result of the logistic regression analysis indicated that daytime activity level was the best predictor of observed depressed mood. The model further suggested that sleep onset latency, total sleep time, and time in bed were also significant predictors of depressed mood. Transition between sleep and wake-fulness was not accepted by the model (see Table 2). Although, in this logistic model daytime activity level was a correlate of depressed mood, the circadian activity amplitude was not. Daytime activity level was expectedly correlated with circadian activity amplitude (r = 0.52, p < 0.01), but was not correlated with time in bed or total sleep time (r = 0.04, NS; r = −0.18, NS, respectively); therefore, no effects of nocturnal sleep on activity were suspected.
Logistic regression assessing the relationship between actigraphic parameters and depressed mood
Conclusions
The results of this investigation suggest that daytime activity level as assessed by wrist actigraphy can be used as an index of depressed mood even in a non-psychiatric sample. The present observation is consistent with previous reports on the association between the percentage of low-level daytime activity and severity of depression [11]. Using activity monitors, other investigators have also reported significant relations between reduced levels of activity and clinical ratings of depression severity among prepubertal children [21] and among college students seeking counselling [22].
Patients with major depression are characterised by significant increase in sleep latency, number of awakenings, and total time awake after sleep onset compared to normal controls [23]; much less is known about the characteristics of sleep in individuals with depressed mood. Baker and colleagues [24] examined sleep and mood differences between healthy premenopausal and perimenopausal women using wrist actigraphy and the Profile of Mood States and found that the sleep disruption of perimenopausal volunteers was significantly greater than that of the premenopausal group. Correlational analyses indicated that sleep and mood changes were significantly related in the perimenopausal group, but not in the premenopausal group. Although data regarding ‘hot flushes’ were not collected, the authors suggested that they might be disturbing the sleep of perimenopausal women, leading to depressed mood.
In the model we developed, depressed mood was significantly correlated with sleep onset latency, wakefulness after sleep onset, total sleep time, and time spent in bed. These findings seem consistent with reports indicating that depressed individuals exhibit significant sleep discontinuity [25] and are characterised by a greater than normal number of nocturnal awakenings [10]. Together, these observations suggest that individuals with mood disturbances may show sleep abnormalities similar to those patients with major depression and that wrist actigraphy might reliably detect them. This is particularly important considering that Mazure and colleagues [26] reported a lack of agreement between subjective reports of depressed patients and objective observations of their sleep/wake patterns. Although subjective reports made by depressed patients are helpful in forming clinical impressions, actigraphy could be used as an adjunct in reliably describing daily activity profile and sleep disturbance naturalistically [11–13].
Several factors might limit the generalisability of the present study. First, our conclusions should be considered tentative until they are replicated in a larger sample. Second, our study addresses the relationship of depressed mood to sleep and psychomotor disturbances based in a non-psychiatric population.
Third, the time criterion for consideration of depressed mood utilised in the Alzheimer's Disease Assessment Scale is at least 1 week, while in the DSM-IV it is 2 or more weeks. However, correlations between depressed mood and sleep disturbance and between depressed mood and psychomotor retardation may even become stronger if a criterion of 2-week duration of symptoms is employed [3].
Despite these limitations, this study replicated previous findings on the correlation between sleep disturbance and depressed mood [24]. Furthermore, this was the first study to explore the relationship between psychomotor retardation and sustained depressed mood in an adult non-psychiatric sample. The findings herein suggest that correlations between depressed mood and sleep abnormalities and between depressed mood and psychomotor disturbances can be found in individuals who do not fulfil the criteria for major depressive disorders. Apart from demonstrating the potential of wrist actigraphy as a research tool, these findings may also have relevant theoretical implications. First, they reinforce the growing body of evidence supporting the view that depression should be considered more as a continuum of depressive symptoms of varying severity and duration [27] than as a set of rigid categories. Second, our results suggest that even ‘subclinical’ depressions may be associated with a certain degree of psychomotor retardation. These findings are in line with Widlöcher's [6] belief that psychomotor retardation is a core behavioural pattern in mood disorders and not a mere secondary symptom reflecting the severity of the depressive episode. Finally, our findings seem to contradict the view that psychomotor disturbance is a specific defining feature of depressive type, being present only in melancholia and absent in non-melancholic depressions [7],[28].
Footnotes
Acknowledgements
This study was supported by a grant from the National Heart Lung and Blood Instittue (HL55983).