PSYCHOSOMATIC MEDICINE, Volume 33, Number 6 Pages 491-497,
November-December 1971

Stress and Neonatal Sleep

Robert N. Emde, MD Robert J. Harmon, MD, David Metcalf, MD, Kenneth L. Koenig, MD and Samuel Wagonfeld, MD

Routine circumcision, done without anesthesia in the newborn nursery was usually followed by prolonged nonrapid eye movement (NREM) sleep. Since this form of sleep has been described as a low point on an arousal continium, we consider its increase to be consistent with a theory of conservation-withdrawal in response to stressful stimulation.

Normal hospital routine requires blood-drawing by lancing the newborn's heel during the first 3 days of life. We observed that this procedure was frequently followed by prolonged sleeping which began with a lengthy, behaviorally inactive sleep period without rapid eye movements (non-REM sleep). This observation was noteworthy because the newborn ordinarily begins a sleep cycle with active, rapid eye movement (REM) sleep (1-4). Curiosity about this reaction to the blood-drawing procedure led us to explore the effects of stressful stimulation on sleeping at this age.

It seemed to us that two different theories offered explanations as to how stressful stimulation would prolong sleep, and that they led to mutually incompatible predictions. One theory was relevant to the general effects of stimulation. It considers REM sleep to serve the primary function of processing and programming information which has been taken in during the wakeful period prior to sleep (5). This theory led us to the prediction of an increase in REM sleep after a large amount of stimulation. A second theory is relevant to the effects of stressful stimulation (6,7). A second theory is relevant to the effects of stressful stimulation (6,7). It postulates two basic physiologic response patterns to such stimulation, one consisting of increased vigilence with mobilization of activity (fight-flight pattern), and the other consisting of a reduction of incoming stimulation by alteration of sensory thresholds with a decline of activity (conservation-withdrawal pattern). Our a assumption was that the immature human infant, who has a limited capacity actively to avoid stressful stimulation, would be likely to respond to such stimulation with the conservation-withdrawal pattern. This led us to predict an increase in non-REM sleep after such stimulation, since this form of infant sleep has been described as a low point on an arousal continuum, a condition in which thresholds to sensory stimulations are high and motoric activity is low (8,9).

Routine hospital circumcision, done without anesthesia, was chosen as a potential stressor which might be expected to produce prolonged bombardment of pain pathways. The Plastibel technic (10) for circumcision was used on all infants studied, is one in which a midline foreskin incision allows a plastic mold to be slipped over the glans penis. A ligature is then tied over the mold which results in a progressive ischemic necrosis of the foreskin during the next several days.

Initial Behavioral Study

Since pilot work verified that prolonged non-REM sleep periods often followed circumcision, we carried out an initial study without polygraphic or other manipulations in which minute-by- minute observations of behavioral states were made. The method of observing behavioral state, along with operational definitions of the states, will be found in a previous publication (3). Infants were observed continuously over a 24-hour period, with individual investigators rotating 4-5 hour observation assignments. Intermittent time sampling for assessing interobserver reliability showed 97% perfect agreement among the twelve states between any two of our four observers. Six normal full-term newborns had circumcisions midway during the observation period. Of the six, 4 evidenced increases in the amount of non-REM sleep of 28%, 72%, and 80%, when the 12 hours before were compared with the twelve hours after the procedure. Increases of this magnitude were remarkable as, in our previous observations of undisturbed infants, we had never seen any comparable values even approaching these. Although all 6 infants spent less time in alert states after circumcision, REM sleep was variable with no consistent increase or decrease. One of the 2 circumcised infants who did not have an increase in non-REM sleep was kept awake by his father for a prolonged period after circumcision. This convinced us of the need for environmental control in a subsequent study. The other infant was unusually alert and wakeful both before and after circumcision and represented a clear exception to the trend of our findings. For purposes of comparison, 3 infants of the same age without circumcision were observed for 24 hours; they showed no increase in non-REM sleep during the second 12 hours.

Polygraphic Study

Method. After this study a second was carried out which employed EEG ash polygraph recordings. Twenty full-term male infants were randomly assigned to 1 of 2 groups - 10 to a circumcision group, and 10 to a control group. One group was circumcised during the period of study and the other group was not. Each group of infants was studied on 2 successive nights, beginning at 24 hours of age. Environmental conditions were standardized with data collected over a 10-hour period, from 10 pm till 8 am. All blood drawing procedures were delayed until after the total period of study. In each instance, observations began with feedings and subsequent feedings were offered in accordance to the infants demands. For the first night's study, conditions were identical in each group, as were the case for the second night's study in the control group. The second night's study of the circumcised group differed only in that infants were circumcised immediately prior to the standardized 10-hour recording period. This experimental design allowed a comparison of postcircumcision nights with first nights of the same infants and also with nights of other infants of the same age who had not been circumcised.

Electrophysiologic recording was done be a 10-channel Model 6 Grass EEG-Polygraph in which four channels were used for recording eye movements three channels for EEG, one channel for respiration. Ongoing behavioral observations were written on the polygraph paper. Minute-by-minute scoring of state was done from the polygraph records by two of the authors, and an agreement study between them on 40 hours of record yielded over 90% agreement on all, sleep, crying and awake states.

Figure 1
Fig.1: Number and duration of non-REM sleep periods; polygraph study Each X represents one non-REM sleep period. Circ, Circumcised.

Results. Results of this study were similar to those of the observational one; of 10 circumcised infants, 8 showed an increase in non-REM or quiet sleep. Increases ranged from 41% to 121% then Night 2 was compared to Night 1. In contrast, in the non-circumcised group, the total amount of non-REM sleep varied little from Night 1 to Night 2; the greatest single percentage increase was less than 3%. Table 1 displays mean data for the 10 infants in each of the two groups. Statistical analysis indicates that increase in non-REM sleep in the postcircumcision subgroup is significant beyond the the 0.01 level of confidence. Analysis of variance was done, using a partially hierarchialized design, with subjects nested within treatments. The interaction F ratio was 16.35; df = 1,18. Although the equal variance assumption was not met for the analysis of variance, the results are consistent with those of the Mann-Whitney U Test applied to the second day-first day differences. It will be noted that the means of crying time are practically identical on Night 2 for circumcised and uncircumcised groups. The lesser amount of awake and drowsy times for the circumcised group on Night 2 is not statistically significant. In 5 of the 8 infants in whom postcircumcision increase in non-REM sleep was noted, there was a corresponding decrease in REM sleep. In the 3 remaining infants, a decrease in the amount of REM sleep was not evident, although a decrease in crying, wakeful and drowsy states was noted.

Postcircumcision increase in non-REM sleep is reflected in several other measures: (a) a shorter latency to the first non-REM sleep period after the feeding which began each observation, (b) an increased number of non-REM sleep periods, and (c) an increased number of extremely long non-REM sleep periods. Shorter latency to non-REM sleep after circumcision was statistically significant at the 0.05 level of confidence. (As in the the previous analysis, a partial hierarchialized design for analysis of variance was used. The F ratio was 7.66; df =1,18.).

Figure 1 illustrates that Night 2 of the circumcision group had twenty more non-REM sleep periods than the control nights of circumcised and noncircumcised groups. It also illustrates that the postcircumcision night was characterized by a wider distribution in the duration of non-REM periods; there were more non-REM which lasted longer than 30 minutes and more which lasted less than 5 minutes.

How long does this effect last? We do not know. Figure 2 suggest a waning in the effect over the 10-hour observation period, but increased values for non-REM sleep did not return to basal levels. The suggested Night 1-Night 2 difference in control group patterns, although not statistically significant, raises a question of a more general effect which is related to the order of recording nights.

Figure 2
Fig. 2 Amount of non-REM sleep over course of each night of observation. Mean values have been calculated for each group using successive 2-hour periods from beginning of observation.

                      Table 1. Mean Values of States
                     Expressed as Percent of Base Time

                               Mean values
                         ------------------------        Mean
                           Night 1      Night 2       difference
     Non-rem sleep
        Noncircumcised      22.15        22.35         + 0.20
        Circumcised         20.37        31.25         +10.88
     REM sleep
        Noncircumcised      50.74        55.16         + 4.42
        Circumcised         48.27        48.37         + 0.10
     Crying time
        Noncircumcised       8.01         5.78         - 2.23
        Circumcised          8.02         5.70         - 2.32
     Awake time
        Noncircumcised       4.70         4.27         - 0.43
        Circumcised          7.87         2.06         - 5.81
     Drowsy Time
        Noncircumcised       4.50         4.12         - 0.38
        Circumcised          7.28         4.91         - 2.39
     Nutritional sucking
     and artifact time
        Noncircumcised       9.90         8.32         - 1.58
        Circumcised          8.19         7.71         - 0.48


Recent reports have indicated that severe physical exercise has the effect of increasing non-REM sleep in rate (11), cats (12) and adolescent humans (13). Is the effect of an increase in non-REM sleep after circumcision primarily due to exercise? At first, anecdotal evidence suggested that it might be. It has been noted in our EEG laboratory that when a technician is learning to apply electrodes (and consequently taking longer to manipulate the infant), sleep often begins with a long behaviorally quiet non-REM period; later as the technician becomes more skillful, this phenomenon is rare. The unskilled manipulation experience could be regarded as one of increased exertion, with crying and consequent fatique; however, it could equally be considered stressful. Further evidence which bears on this question was uncovered in a review of data from the current polygraphic study. The circumcision procedure added a mean of 15.6 minutes to the night's observation; much of this time was spent in vigorous crying with considerable exercise. For purposes of comparison, we found two instances from our ongoing naturalistic studies, in which newborn babies cried for long periods (47 and 87 minutes) in the absence of circumcision or other stressful procedures. We were able to compare intervals with of at least 6 hours before and after the crying period for each infant. We found no increase in non-REM sleep after crying and no decrease in latency to the first non-REM sleep period. It therefore seem unlikely that exercise is the major factor in producing this effect.

Another alternative to the conservation-withdrawal explanation of the postcircumcision effect is suggested by Oswald (14), who postulates that non-REM sleep assists in the processes of growth and repair of body tissues outside of the central nervous system. A test of this explanation may be found in an experimental design comparing circumcision effects on infants who had local anesthesia with infants who did not. If the effect of increased non-REM sleep is a function of bombarded pain pathways and resultant physiologic stress, the anesthetic condition should eliminate it. However, if this effect is a result of processes of tissue breakdown and repair, the effect would remain.

The fact that non-REM sleep declined in some infants invites a comparison with the first night effect in adults, in which initial laboratory sleeping results in a lesser amount of REM sleep than occurs in subsequent laboratory nights (15). Although such a comparison is limited because the effect has not been documented in human infants, the postcircumcision effect would appear to be different, since there was not an increase in the amount of quiet non-REM sleep.

In conclusion, the information processing theory of REM sleep led us to a false prediction in this study, while the conservation-withdrawal theory led us to an accurate one. This would seem to provide further confirmatory evidence for the usefulness of the latest theory which has been employed in explaining the experimental production of pathologic states in infant pigtail monkeys (Macaca nemestrina) (16). Current research is aimed at the endocrinologic basis for the phenomenon of increased non-REM sleep after circumcision. This phenomenon may be related to some long-term aspects of infant development; these are being investigated along theoretical lines proposed elsewhere (17).


Prolonged periods of quiet non-REM sleep sometimes followed routine blood drawing in newborns. This observation led us to study the effects of stressful stimulation on sleep patterns at this age.

Routine hospital circumcision, done without anesthesia, was chosen as a potential stressor which might be expected to produce prolonged bombardment of pain pathways. Two studies, one without polygraphic manipulation and one with EEG and polygraphic manipulation and one with EEG and polygraphic recording, resulted in similar findings. Circumcision was usually followed by prolonged, non-REM sleep. Effects of circumcision were demonstrable in terms of an increase in the amount of non-REM sleep (p<0.01) and a decrease in latency to the onset of non-REM sleep (P<0.05). Infants were used as their own controls and were compared with non-circumcised males for statistical analysis. Postcircumcision increase in non-REM sleep was also reflected in an increased total number of non-REM sleep periods and an increased number of extremely long non-REM sleep periods.

These results are interpreted to be consistent with a theory of conservation-withdrawal in response to stressful stimulation.


  1. Roffwarg HP, Muzio JN, Dement WC: Ontogenetic development of the human sleep-dream cycle. Science 152:604-619, 1966. [Abstract]
  2. Emde RN, Koenig KL: Neonatal smiling and rapid eye movement states. J Amer Acad Child Psychiat 8:57-67, 1969. [PubMed]
  3. Emde RN, Koenig Kl: Neonatal smiling, frowning and rapid eye movement states II. Sleep cycle study. J Amer Acad Child Psychiatr 8:637-656, 1969. [PubMed]
  4. Emde RN, Metcalf DR: An electroencephalographic study of behavioral rapid eye movement states in the human newborn. J Nerv Ment Dis 150:376-386, 1970. [PubMed]
  5. Dewan, EM: The programming (p) hypothesis for REMs AFCRL-69-0298:388, 1969.
  6. Engel, GL: Anxiety and depression withdrawal: the primary effects of unpleasure. In J Psychoanal 43:89-97, 1962. [Abstract]
  7. Engel GL: Psychological Development in Health and Disease. Philadelphia, WB Saunders, 1962, pp 384-388.
  8. Stern E, Parmelee AH, Akiyama Y, et al. Sleep cycle characteristics in infants. Pediatrics 43:65-70, 1969. [Abstract]
  9. Wolff PH: The causes, controls, and organization of behavior in the neonate. Psychological Issues. Vol 17. New York, International Universities Press., 1966. [PubMed]
  10. Barrie, H. Huntington PJ, Gough M: The Plastibell technique for circumcision. Brit Med J 2:273-275, 1965. [Full Text]
  11. Matsumoto J, Nishisho T, Suto T, et al: Influence of fatigue on sleep. Nature 218:177-178, 1968.
  12. Hobson, JA: Sleep after exercise. Science 162:1503-1505, 1968. [Abstract]
  13. Baekland F, Lasky R: Exercise and sleep patterns in college athletes. Pecept Motor Skills 23:1203-1207, 1966. [PubMed]
  14. Oswald, I: Sleep, the great restorer. New Scientist 46:169-172, 1970.
  15. Agnew HW, Webb WE, Williams RL: The first night effect: an EEG study of sleep. Psychophysiology 2:263-266, 1966. [Abstract]
  16. Kaufman IC, Rosenblum LA: Depression in infant monkeys separated from their mothers. Science 155:1030-1031. [Abstract]
  17. Spitz RA, Emde RN, Metcalf DR: Further propotypes of ego formation: a working paper from a research project on early development. Psychoanal Stud Child 25:417-441, 1970.

From the Department of Psychiatry, University of Colorado, School of Medicine, Denver, Colo.
Supported by US Public Health Service Grant MH-HD 15753 and National Institute of Mental Health Research Scientist Development Awards K3-MH-36808 and K5-MH-40275.
The authors wish to thank Donald W. Stilson for statistical consultation.
Address for reprint requests: Robert N. Emde, MD, Department of Psychiatry, University of Colorado Medical Center, 4200 East Ninth ave, Denver, Colo 80220.

(File revised 22 July 2008)