674
Electroencephalography and Clinical Neurophysiology, 1978, 4 4 : 6 7 4 - - 6 7 6 © Elsevier/North-Holland Scientific Publishers Ltd.
Technical contribution WRIST ACTIGRAPHIC MEASURES OF SLEEP AND RHYTHMS* D.F. KRIPKE**, D.J. MULLANEY, S. MESSIN and V.G. WYBORNEY
Department of Psychiatry, University of California, San Diego and San Diego Veterans Administration Hospital, San Diego, Calif. 92161 (U.S.A.) (Accepted for publication : October 7, 1977) Recently, Kupfer (1972), Foster (1972) and colleagues have described the use of an actigraphic telemetry system for quantifying both sleep and biologic rhythms. They have reported surprisingly high correlations between wrist activity and wakefulness as determined by EEG -- 0.84 and 0.88 in separate studies (Kupfer et al. 1972; Kupfer and Foster 1973). Encouraged by their results, we have developed a more flexible system and have validated its usefulness in distinguishing sleep and wakefulness.
Method and Materials An activity transducer was developed by soldering a small steel nut off-center onto a 5-ram length of spring-like EEG pen wire, and clamping the other end of the wire against a piezo-ceramic element (Messin et al. 1975). Because the weight of the nut is offcenter, the piezo-ceramic element is excited when the transducer is moved in any direction, although the voltage is not linearly related to motion or acceleration. Two 10-m~ resistors complete the transducer, which is packaged in a small acrylic box m o u n ted on a watch band and connected to a Medilog tape recorder*** (Fig. 1). The analog voltage is recorded on small C-120 cassettes which record continuously for 24 h. For precise time measurements, a time code should be recorded on an additional channel. We have successfully employed the voltage supplied to the minutes digit display of an LCD watch as a time code. Subjects wear the activity transducer night and * Supported in part by a Research Scientist Development Award (1 K02 M H 0 0 1 1 7 ) f r o m the National Institute of Mental Health, by NIMH MH-21350 and by the Medical Research Service of the Veterans Administration. * * R e p r i n t requests should be sent to: Dr. D.F. Kripke, Veterans Administration Hospital, 3350 La Jolla Village Drive, San Diego, Calif. 92161, U.S.A. *** Ambulatory Monitoring, Inc., 731 Saw Mill River Road, Ardsley, NY 10502.
day (it has proven relatively shower-proof). The tape recorder is worn on a belt or placed by the bed at night. Activity analogs can be replayed on a Grass 78B polygraph at 15 ram/rain so that a 24-h tracing may be displayed in a concise 82-page record (2460 cm). In addition to activity, the Medilog recorder has been used to simultaneously record parieto-occipital EEG, horizontal eye movements and submentat EMG on 3 additional channels, using methods similar to those described by Wilkinson et al. (1973), Ives and Woods (1975) and Wilkinson and Mullaney (1976). These 3 channels were then replayed to obtain standard polygraphic sleep records (15 ram/ sec). Results The activity transducer has been well tolerated and reliable in several hundred 24-h recordings of psychiatric patients and normal subjects. Ambulatory subjects have been recorded for as long as 30 days continuously while they were at home and at work. Subjects moved their wrists almost continuously when awake, at least once a minute, but only brief bursts of movement were seen during sleep (Fig. 2). About 9 h of recording including all sleep were analysed for 5 normal subjects who slept at home. EEG-EOG-EMG recordings were scored in l-rain epochs for Wake (including Wake and Movement Time) or Sleep (Stages 1, 2, 3, 4 and REM) according to Rechtschaffen and Kales (1968) criteria. Each minute of actigraphic tracing was scored Wake or Sleep using the best guess which could be derived. Separate actigraphic and standard sleep recordings were scored independently by separate blind raters. Table I presents the results of these blind independent scorings for the 5 subjects. The actigraphic measure gave highly reliable estimates of the minutes of sleep, Total Sleep Period and the minutes of Wake Time Within Sleep. Actigraphic and EEG estimates of the m o m e n t of sleep onset agreed within 2 rain for 4 out of 5 subjects. The 5th subject had a 46-rain sleep latency by EEG estimations, but the sleep latency estimate by actigraph was 61 rain.
WRIST ACTIGRAPHIC MEASURES OF SLEEP AND RHYTHMS
675
\ METAL WEIGHT i
0
CRYSTAL
'~,,
I
SPRING WIRE
Fig. 1. The actigraphic transducer and analog recorder.
Discussion Our findings c o n f i r m earlier reports of K u p f e r ' s group t h a t wrist activity m e a s u r e m e n t s m a y be sufficient for reliable estimations o f sleep duration. Piezo-electric transducers m a y be m o r e omni-directional and m o r e sensitive than those previously described. C o n t i n u o u s analog activity recordings have - AWAKE
_L -r
SLEEP PERIOD
_L -I-
,
TABLE I Comparison o f t w o m e t h o d s o f sleep m e a s u r e m e n t
Sleep period
AWAKE -
~ e e p time
I
1 hr.
I
Fig. 2. Compressed 10-h record o f an integrated actigraphic analog, showing the contrast in activity b e t w e e n sleep and wake time.
Wake time Within sleep
Polygraph
Actigraph
436 min 442 497 466 512
433 min 446 485 454 528
r = 0.954 t = 5.49 P < 0.01
434 min 348 442 462 502
410 min 345 439 453 473
r = 0.982 t = 9.01 P < 0.005
94 min 54 4 10 2
100 min 46 1 55 23
r = 0.851 t = 2.80 P < 0.05
676
D.F. KRIPKE ET AL.
several hundred times the information content of mean 5-min counts which have been used previously. Thus, the current technique seems to present improved scoring reliability. A major advantage of tape recording is its applicability to subjects outside the hospital. Wrist actigraphs may permit large-scale pharmacologic and epedemiologic studies of sleep which would be impractical with EEG methods or activity telemetries. Wrist actigraphs may also be useful for long-term circadian rhythm studies in humans analogous to chronobiologic studies of animal activity (Pittindrigh 1974).
Summary With a piezo-electric activity transducer, wrist activity recordings were made simultaneously with EEG, EOG and EMG recordings. Wrist activity alone was used to estimate Total Sleep Time and blind independent scoring of sleep and wakefulness by EEG-EOG-EMG was done. Correlations between the two methods were termined for minutes of Sleep (r = 0.98), Total Sleep Period (r = 0.95) and minutes of Wake Time Within Sleep (r = 0.85). Continuous wrist activity recordings may thus provide inexpensive but very accurate estimates of Sleep Time.
l~sum~ Mesures a c t i g r a p h i q u e s sommeil et rythmes
du
poignet
au
cours d u
Au moyen d ' u n capteur piezo-~lectrique d'activit~, des enregistrements de l'activit~ du poignet ont ~t~ effectu~s en m~me temps que des enregistrements EEG, EOG et EMG. L'activit~ du poignet seule est utilisde pour l'estimation du temps de sommeil total, et des codages ind~pendants du sommeil et de la veiUe ont dt$ rdalisds a l'aveugle au moyen de l'ensemble EEG, EOG, EMG. Les correlations entre ces deux mgthodes ont ~t~ dSterminges en ce qui con-
cerne les minutes de sommeil (r ffi 0.98), la p~riode de sommeil total (r = 0.95) et les minutes de temps d'eveil a l'interieur du sommeil (r = 0.85). L'enregistrement continual de l'activit~ du poignet peut ainsi fournir des estimations peu co~teuses mais tr~s pr~cises du temps de sommeil.
References
Foster, F.G., Kupfer, D.J., Weiss, G., Lipponen, U., McPartland, R. and Delgado, J.R. Mobility recording and cycle research in neuropsychiatry. J. Interdis. Cycle Res., 1972, 3(1): 61--72. Ives, J.R. and Woods, J.F. 4-Channel 24 hour cassette recorder for longterm EEG monitoring of ambulatory patients. Electroenceph. clin. Neurophysiol., 1975, 39: 88--92. Kupfer, D.F., Detre, T.P., Foster, F.G., Tucker, G. and Delgado, J. The application of Delgado's telemetric mobility recorder for h u m a n sleep studies. Behav. Biol., 1972, 7: 585--590. Kupfer, D.J. and Foster, F.G. Sleep and activity in a psychotic depression. J. nerv. ment. Dis., 1973, 156: 341--348. Messin, S., Kripke, D.F., Atkinson, M. and Forney, E. An implantable eye movement transducer. Electroenceph, clin. Neurophysiol., 1975, 38: 643--644. Pittendrigh, C.S. Circadian oscillations in cells and the circadian organization of multicellular systems. In F.O. Schmitt and F.G. Worden (Eds.), The Neurosciences: Third Study Program. MIT Press, Cambridge, Mass., 1974: 437--458. Rechtschaffen, A. and Kales, A. Manual of Standardized Terminology, Techniques and Scoring System for Sleep Stages of Human Subjects. NIH Pub. #204, Washington, Superintendent of Documents, Book 1--62, 1968. Wilkinson, R.T., Herbert, M. and Branton, P.A. A portable EEG-Sleep recorder of pocket size. Sleep Res., 1973: 2: 182. Wilkinson, R.T. and Mullaney, D. Electroencephalogram recording of sleep in the home. Postgrad. Med. J., 1976, 52(7): 92--96.
Electroencephalography and Clinical Neurophysiology, 1978, 4 4 : 6 7 4 - - 6 7 6 © Elsevier/North-Holland Scientific Publishers Ltd.
Technical contribution WRIST ACTIGRAPHIC MEASURES OF SLEEP AND RHYTHMS* D.F. KRIPKE**, D.J. MULLANEY, S. MESSIN and V.G. WYBORNEY
Department of Psychiatry, University of California, San Diego and San Diego Veterans Administration Hospital, San Diego, Calif. 92161 (U.S.A.) (Accepted for publication : October 7, 1977) Recently, Kupfer (1972), Foster (1972) and colleagues have described the use of an actigraphic telemetry system for quantifying both sleep and biologic rhythms. They have reported surprisingly high correlations between wrist activity and wakefulness as determined by EEG -- 0.84 and 0.88 in separate studies (Kupfer et al. 1972; Kupfer and Foster 1973). Encouraged by their results, we have developed a more flexible system and have validated its usefulness in distinguishing sleep and wakefulness.
Method and Materials An activity transducer was developed by soldering a small steel nut off-center onto a 5-ram length of spring-like EEG pen wire, and clamping the other end of the wire against a piezo-ceramic element (Messin et al. 1975). Because the weight of the nut is offcenter, the piezo-ceramic element is excited when the transducer is moved in any direction, although the voltage is not linearly related to motion or acceleration. Two 10-m~ resistors complete the transducer, which is packaged in a small acrylic box m o u n ted on a watch band and connected to a Medilog tape recorder*** (Fig. 1). The analog voltage is recorded on small C-120 cassettes which record continuously for 24 h. For precise time measurements, a time code should be recorded on an additional channel. We have successfully employed the voltage supplied to the minutes digit display of an LCD watch as a time code. Subjects wear the activity transducer night and * Supported in part by a Research Scientist Development Award (1 K02 M H 0 0 1 1 7 ) f r o m the National Institute of Mental Health, by NIMH MH-21350 and by the Medical Research Service of the Veterans Administration. * * R e p r i n t requests should be sent to: Dr. D.F. Kripke, Veterans Administration Hospital, 3350 La Jolla Village Drive, San Diego, Calif. 92161, U.S.A. *** Ambulatory Monitoring, Inc., 731 Saw Mill River Road, Ardsley, NY 10502.
day (it has proven relatively shower-proof). The tape recorder is worn on a belt or placed by the bed at night. Activity analogs can be replayed on a Grass 78B polygraph at 15 ram/rain so that a 24-h tracing may be displayed in a concise 82-page record (2460 cm). In addition to activity, the Medilog recorder has been used to simultaneously record parieto-occipital EEG, horizontal eye movements and submentat EMG on 3 additional channels, using methods similar to those described by Wilkinson et al. (1973), Ives and Woods (1975) and Wilkinson and Mullaney (1976). These 3 channels were then replayed to obtain standard polygraphic sleep records (15 ram/ sec). Results The activity transducer has been well tolerated and reliable in several hundred 24-h recordings of psychiatric patients and normal subjects. Ambulatory subjects have been recorded for as long as 30 days continuously while they were at home and at work. Subjects moved their wrists almost continuously when awake, at least once a minute, but only brief bursts of movement were seen during sleep (Fig. 2). About 9 h of recording including all sleep were analysed for 5 normal subjects who slept at home. EEG-EOG-EMG recordings were scored in l-rain epochs for Wake (including Wake and Movement Time) or Sleep (Stages 1, 2, 3, 4 and REM) according to Rechtschaffen and Kales (1968) criteria. Each minute of actigraphic tracing was scored Wake or Sleep using the best guess which could be derived. Separate actigraphic and standard sleep recordings were scored independently by separate blind raters. Table I presents the results of these blind independent scorings for the 5 subjects. The actigraphic measure gave highly reliable estimates of the minutes of sleep, Total Sleep Period and the minutes of Wake Time Within Sleep. Actigraphic and EEG estimates of the m o m e n t of sleep onset agreed within 2 rain for 4 out of 5 subjects. The 5th subject had a 46-rain sleep latency by EEG estimations, but the sleep latency estimate by actigraph was 61 rain.
WRIST ACTIGRAPHIC MEASURES OF SLEEP AND RHYTHMS
675
\ METAL WEIGHT i
0
CRYSTAL
'~,,
I
SPRING WIRE
Fig. 1. The actigraphic transducer and analog recorder.
Discussion Our findings c o n f i r m earlier reports of K u p f e r ' s group t h a t wrist activity m e a s u r e m e n t s m a y be sufficient for reliable estimations o f sleep duration. Piezo-electric transducers m a y be m o r e omni-directional and m o r e sensitive than those previously described. C o n t i n u o u s analog activity recordings have - AWAKE
_L -r
SLEEP PERIOD
_L -I-
,
TABLE I Comparison o f t w o m e t h o d s o f sleep m e a s u r e m e n t
Sleep period
AWAKE -
~ e e p time
I
1 hr.
I
Fig. 2. Compressed 10-h record o f an integrated actigraphic analog, showing the contrast in activity b e t w e e n sleep and wake time.
Wake time Within sleep
Polygraph
Actigraph
436 min 442 497 466 512
433 min 446 485 454 528
r = 0.954 t = 5.49 P < 0.01
434 min 348 442 462 502
410 min 345 439 453 473
r = 0.982 t = 9.01 P < 0.005
94 min 54 4 10 2
100 min 46 1 55 23
r = 0.851 t = 2.80 P < 0.05
676
D.F. KRIPKE ET AL.
several hundred times the information content of mean 5-min counts which have been used previously. Thus, the current technique seems to present improved scoring reliability. A major advantage of tape recording is its applicability to subjects outside the hospital. Wrist actigraphs may permit large-scale pharmacologic and epedemiologic studies of sleep which would be impractical with EEG methods or activity telemetries. Wrist actigraphs may also be useful for long-term circadian rhythm studies in humans analogous to chronobiologic studies of animal activity (Pittindrigh 1974).
Summary With a piezo-electric activity transducer, wrist activity recordings were made simultaneously with EEG, EOG and EMG recordings. Wrist activity alone was used to estimate Total Sleep Time and blind independent scoring of sleep and wakefulness by EEG-EOG-EMG was done. Correlations between the two methods were termined for minutes of Sleep (r = 0.98), Total Sleep Period (r = 0.95) and minutes of Wake Time Within Sleep (r = 0.85). Continuous wrist activity recordings may thus provide inexpensive but very accurate estimates of Sleep Time.
l~sum~ Mesures a c t i g r a p h i q u e s sommeil et rythmes
du
poignet
au
cours d u
Au moyen d ' u n capteur piezo-~lectrique d'activit~, des enregistrements de l'activit~ du poignet ont ~t~ effectu~s en m~me temps que des enregistrements EEG, EOG et EMG. L'activit~ du poignet seule est utilisde pour l'estimation du temps de sommeil total, et des codages ind~pendants du sommeil et de la veiUe ont dt$ rdalisds a l'aveugle au moyen de l'ensemble EEG, EOG, EMG. Les correlations entre ces deux mgthodes ont ~t~ dSterminges en ce qui con-
cerne les minutes de sommeil (r ffi 0.98), la p~riode de sommeil total (r = 0.95) et les minutes de temps d'eveil a l'interieur du sommeil (r = 0.85). L'enregistrement continual de l'activit~ du poignet peut ainsi fournir des estimations peu co~teuses mais tr~s pr~cises du temps de sommeil.
References
Foster, F.G., Kupfer, D.J., Weiss, G., Lipponen, U., McPartland, R. and Delgado, J.R. Mobility recording and cycle research in neuropsychiatry. J. Interdis. Cycle Res., 1972, 3(1): 61--72. Ives, J.R. and Woods, J.F. 4-Channel 24 hour cassette recorder for longterm EEG monitoring of ambulatory patients. Electroenceph. clin. Neurophysiol., 1975, 39: 88--92. Kupfer, D.F., Detre, T.P., Foster, F.G., Tucker, G. and Delgado, J. The application of Delgado's telemetric mobility recorder for h u m a n sleep studies. Behav. Biol., 1972, 7: 585--590. Kupfer, D.J. and Foster, F.G. Sleep and activity in a psychotic depression. J. nerv. ment. Dis., 1973, 156: 341--348. Messin, S., Kripke, D.F., Atkinson, M. and Forney, E. An implantable eye movement transducer. Electroenceph, clin. Neurophysiol., 1975, 38: 643--644. Pittendrigh, C.S. Circadian oscillations in cells and the circadian organization of multicellular systems. In F.O. Schmitt and F.G. Worden (Eds.), The Neurosciences: Third Study Program. MIT Press, Cambridge, Mass., 1974: 437--458. Rechtschaffen, A. and Kales, A. Manual of Standardized Terminology, Techniques and Scoring System for Sleep Stages of Human Subjects. NIH Pub. #204, Washington, Superintendent of Documents, Book 1--62, 1968. Wilkinson, R.T., Herbert, M. and Branton, P.A. A portable EEG-Sleep recorder of pocket size. Sleep Res., 1973: 2: 182. Wilkinson, R.T. and Mullaney, D. Electroencephalogram recording of sleep in the home. Postgrad. Med. J., 1976, 52(7): 92--96.
