EARLY HUMAN DEVELOPMENT, Volume 7, Number 4: Pages 367-374,
December 1982.


II. Effects upon mother-infant interaction

Richard E. Marshall, Fran L. Porter, Ann G. Rogers, JoAnn Moore, Barbara Anderson and Stuart B. Boxerman

Washington University School of Medicine, St. Louis, Missouri, U.S.A.

Accepted for publication 21 October 1982


The effects of circumcision upon mother-infant interaction were examined in an observational study of 59 mother-infant pairs during hospital feedings using a specifically designed mother-infant interaction observation system that examined 43 discreet behaviours relating to feeding, gaze, facial expression, vocalizations and touch. The experimental group was circumcised after the second feeding and the control group after the fourth feeding.
Analysis revealed no major differences between the experimental and control groups. Yet, different trends between the two groups were observed regarding two variables shortly after surgery. These differences disappeared by 24 h post-operatively. Differences relating to the frequency of feeding intervals and infant availability scores. The study also revealed a surprising limited repertoire of behavior exhibited by both the mother and infant during feeding sessions.
Our data suggest that circumcision has brief and transitory effects on mother-infant interactions observed during hospital feeding sessions, the only time mothers who are not rooming in have to be with their infants.

circumcision; mother-infant interaction; infant availability; feeding patterns


There are limited ways in which it is possible to study the effects of pain on healthy infants. Circumcision provides one such opportunity. Although approximately one million newborn males are routinely circumcised each year, the short term consequences of this procedure are poorly understood2,3. In a previous study5 we examined infants before and after circumcision using the Brazelton Neonatal Behavioral Assessment Scale (NBAS)1, a series of stimuli designed to elicit measured response from infants. We found that 4 h after circumcision, 90% of the infants reacted to stimuli differently than they had pre-operatively. Approximately half of the infants became more active, agitated and fussy following circumcision, and half more subdued, drowsy or sleepy. Within 24 h, however, most had reverted to their pre-circumcision responses.

In the current study, we wished to examine the impact of circumcision upon the mother-child interaction as it occurs in an naturalistic hospital setting for a clinic population. We chose the feeding situation for a variety of reasons. (1) For those mothers who do not elect to have their babies room with them (rooming-in), it is the only predictable and regular time infants have with their mothers. (2) The feeding time is an actual situation in which mothers interact with their babies; it, therefore, is not a manipulative stressful situation like the Brazelton exam. We choose clinic patients since circumcisions on private patients are done by private obstetricians at a time convenient to their schedules, thus rendering controlled observations on these patients difficult.

Materials and Methods

59 newborn infants were randomly assigned to either an experimental (early circumcision) or control (delayed circumcision) group and were observed during four hospital feeding sessions using a specifically designed mother-infant interaction observation system 6. Each mother-infant dyad was observed by one of two observers during the first 15 min of four hospital feeding sessions occuring in a 24-h period. Infants in the experimental groups (n=27) were circumcised on the 2nd day of life, between observations 2 and 3, and those in the control group (n=32) on the 3rd day of life after the completion of four observations (Table I). This protocol allowed us to compare the early circumcision group with a similar group without surgical intervention. All clinic mothers are discharged before noon on the third day of life, making further observation of the delayed groups impossible.


Observation schedule

Early circumcision                   Delayed circumcision
OB1  9:00 a.m. 2nd day               OB1 9:00 a.m. 2nd day
OB2  1:00 p.m. 2nd day               OB2 1:00 p.m. 2nd day

     Circumcision                        Circumcision
OB3  5:00 p.m 2nd day                OB3 5:00 p.m. 2nd day
OB4  9:00 a.m.3rd day                OB4 9:00 a.m. 3rd day                                         


Data Collection form


Name:       Feeding      Gaze         Facial     Vocal        Touch
Date:       Codes
Observer:   Mother       Mother       Mother     Mother       Mother
            1-none       1-not at I    1-neutral  1-none       0-not hold
            2-presents   2-looks,     2-frown    2-indistinct 1-none
            3-continues    glances    3-bright   3-rhythmic   2-fingertip
            4-removes      away       or smile   4-speech     3-palm
            5-wipes      3-looks at I 4-animated              4-kiss,

            Infant       Infant       Infant     Infant       Infant
            1-none       1-eyes       1-neutral  1-none       1-none
            2-not accept closed       2-frown    2-fuss,cry   2-startle
            3-accept     2-eyes open  3-bright,  3-audible    3-mold, root
            4-sucking    3-focus on     smile      signal       cling
            5-spits        M's face   4-yawn     4-burp       4-averts head

All males whose mothers were clinic patients on the Washington University OB-GYN delivery service were considered eligible candidates for our study. Mothers and infants were screened to meet specific criteria4,5 including vaginal delivery with mild or no anesthesia, term gestation, normal physical health, and bottle feeding without rooming-in. The mothers were told that their infants would be circumcised on either day 2 or 3. In an effort to control for maternal behaviors elicited by knowledge that their babies had been circumcised, mothers were requested not to look at their infants genitals or ask nurses whether their babies had been circumcised. Nurses reported to us that mothers did not ask about circumcision status of their infants. Babies whose mothers consented to experimental participation were randomly assigned to either the experimental or control group. Circumcisions were performed using a Gomco clamp and a circumcision board (Olympic Circumstraints, Olympic Medical, Seattle, WA) that fixed the arms and legs.

A sample data sheet is shown in Table II. A time-sampling strategy was used to make observations in the mother's hospital room. Over an approximate 13-min period forty 5-s observations were made. Blocks of 2 consecutive 5-s observations were followed by 30 s for recording. Time intervals were determined by pre-recorded signals audible only to the observer from a cassette recorder. Inter-observer reliability was 83% based on the results of observations made simultaneously by 2 independent observers periodically throughout the study.

Five behavioral domains were observed and recorded for both mothers and infants during each 5-s interval. These included feeding, gaze, facial expression, vocalizations and touch. 43 distinct codes (23 maternal; 20 infant) defined all observed behaviors. These are itemized on the data sheet (Table II). In addition, infant states, as defined in the NBAS, were recorded for each 5-s observation interval. Finally, in addition to the quantitatively scored data, each observer added a brief paragraph subjectively describing the interaction patterns of the dyad.

Because our previous study suggested that infant state behavior was altered by circumcision we wanted to determine whether differences in infant state would be reflected in a non-stressful feeding situation as well. To summarize differences in infant state, we developed an index of infant availability which was based on the assumption that infant in a calm alert state (state 4) were more available for social interaction than when they were sleeping (states 1,2), drowsy (state 3), fussy (state 5), or crying (state 6). The availability index, which weights states according to heir potential availability, is shown in Table III.


Characteristics of the early and delayed circumcision groups, including type of maternal anesthesia, race, marital status, and infant gestational age were compared using Fisher's exact test. Other maternal and infant factors, such as gravidity, education, Apgar scores, birth weight and length were compared using two-sample t-tests. No significant differences were found between the two groups for any factors using a significance level of 5%.


Infant availability scale

Infant      Score   X    Intervals occurred X 40
1,2         0            Unavailable-lowest score            = 0
3,4,5       1            Partially available-average score   =40
4           2            Fully available-highest score       =80

Although the feeding situation is the primary and sometimes only time that mothers who are not rooming-in are together with their infants in a hospital, we found very little indication of any robust dyadic interaction. To get an overview of the early behavioral repertoire of mothers and infants, we combined data from all subjects in both early and delayed circumcision groups across all observations.

We found that mothers primarily held their infants in a cradled position (73%), looked at their infant's faces (83%) with a neutral facial expression (89%), and displayed little touching (25%) or vocalizing (22%). Overall, mothers attempted to feed their infants 62% of the time. Concurrently, the infants eyes were typically closed (71%), had neutral facial expressions (91%), did little vocalizing (8%), clinging (13%), or even feeding (40%).

To further examine the behavioral repertoire of mothers and babies the frequency that each behavior in a domain was observed to occur in combination with other domain behaviors was calculated. Behavioral combinations occurring less than 10 times across all observations were omitted. These results illustrated again how limited the behavioral repertoire of mothers and infants is during a feeding situation as well as how infrequently dyadic social interaction occurs. Only 3 combinations of 2 maternal behaviors (out of 2000 theoretically possible) account for 49% of all dyadic interaction. The combinations occuring with the highest frequencies reflected only behaviors regarding feeding and gaze; all other combinations occurred with lesser frequencies and behaviors in the domains of facial expression, vocalization, and touch seldom occurred.

To summarize, mothers predominantly looked at their infants while holding and trying to feed them. Babies either fed with their eyes closed or did not feed at all. The average number of intervals in which infant feeding occurred did increase across time, rising from 59% at observation 1 to 87% at observation 4. This trend of the total population is due, most probably, to both the increased requirements and skills of the maturing infant.

Although an analysis of most maternal and infant behavior revealed no significant differences between experimental and control groups, results of interest appeared for 2 variables: the number of intervals in which feeding occurred (Fig. 1) and infant availability (Fig. 2). When the frequency of uninterrupted infant feeding was compared for the 2 groups for each observation we found that the experimental showed a mean change of -0.05 between observation 2 before circumcision and observation 3 (after circumcision). During this same period the control group showed a mean change of 2.73, clearly demonstrating the lack of impact of surgical intervention (Wilcoxon P<0.05). We found a similar leveling off of experimental maternal feeding behavior (X change = 0.07 compared to X change = 3.13 for the control group mothers) although this difference was significant only at the 0.07 level (Wilcoxon).

[Two line charts plotted on x-y axis. Since line charts cannot be presented here CIRP presents data in tabular form.]

Mother Continues

         Control      Experimental
OB 1     25%          32%
OB 2     31%          37%
OB 3     41%          40%
OB 4     48%          40%

Infant sucks

         Control      Experimental
OB 1     30%          35%
OB 2     35%          41%
OB 3     42%          41%
OB 4     50%          42%
Fig. 1. Percent intervals containing feeding behavior for mothers (top) and infants (bottom) across observations for experimental (circumcised) and control (non-circumcised) groups.

[Line chart on plotted on x-y axis. Since line charts cannot be presented here, CIRP presents data in tabular form]

         Control      Experimental
OB 1     44%          35%
OB 2     40%          36%
OB 3     45%          39%
OB 4     54%          50%
Fig. 2. Mean infant availability scores (see Methods) across observations for experimental (circumcised) and control (non-circumcised) groups.

We examined whether this difference in feeding behavior might be related to differences in the state behavior of the infants in the 2 groups. As seen in Figure 2 between the 2nd observation (before circumcision) and the 3rd observation (after circumcision) than did the control group. Although this did not reach significant levels (Wilcoxon P<0.09) it does reflect a different trend between the two groups, specifically that the circumcised infants displayed less state 4 behavior than the control groups.


Our previous work demonstrated that infants who are circumcised undergo behavioral changes detected after circumcision by using the NBAS. In this study, we chose to examine behavior in a caregiving clinical environment by evaluating the impact of circumcision on mother-infant interaction. The feeding session revealed a limited repertoire of maternal and newborn interactive behaviors; the mother tried to feed her baby who was wrapped in a blanket. The fact that observers were in the room may well have influenced our results. Mothers may have been inhibited by the observer's presence but it could also be suggested that mothers might have tried harder to socialize with their infants since they were `on display'. Clearly, however the behavioral changes were much less dramatic than those observed in response to NBAS examinations following circumcision.


Availability scores (X)
Observation        1         2          3           4
Experimental       35.41     37.11      37.41       52.26
Control            42.53     38.97      45.41       54.06

We did find, however, that immediately following circumcision there were differences in the feeding patterns between the two groups. The experimental group exhibited fewer intervals of uninterrupted feeding than did the control group. Similar patterns, although not at highly significant levels, were demonstrated by the experimental group mothers. Although the volume of formula consumed was not monitored, subjective descriptions support the observations that the infants who were circumcised sucked on the bottles harder, faster and more concertedly. Their comments revealed that certain infants appeared to be intensely focused on feeding at observation 3 and seemed to `tune out' external distractions. Subsequently we determined that over 85% of these particular individuals were in the experimental group. This explanation would lend support to the different trends found in availability scores between observations 2 and 3 for the 2 groups. If infants in the experimental group did concentrate more on feeding, they would be less available for social interaction.

[CIRP Note: The babies in this 1982 study were fed breastmilk substitute (formula) from bottles. Today, we know that mother's milk is superior to breastmilk substitute and breastfeeding is encouraged. Suckling at the breast is more difficult for newborns than taking breastmilk substitute from a bottle. When stressed, exhausted, and traumatized from a circumcision, newborn babies may be unable to manage to suckle at the breast.]

It is certainly possible that there are effects of circumcision that can affect mother-infant interaction during a non-feeding situation which are not precluded on the basis of our data. It is also possible that circumcision might affect interaction between infants and their primary caregivers, the nurses. If the subjects in the delayed circumcision group had been available for more extensive observations perhaps further differences between the two groups might have emerged. Further investigations are, indeed, indicated.

The study revealed that in a hospital setting with restricted opportunity for mother-infant interaction both mothers and infants limit their behaviors to feeding-related events when they are together. A stressful, painful event such as circumcision, however, still appears to affect the feeding patterns of infants who have recently undergone circumcision. Despite differences between control and experimental infants shortly after surgery, by 24 h post-operatively no significant differences were observed between the groups. The behavioral effects of circumcision in the present study were immediate but brief. This should be comforting information to those who provide care for newborns and for their parents.

[CIRP Note: Long-term behavioral changes since have been documented in circumcised boys. See Effect of neonatal circumcision on pain response during subsequent routine vaccination.]


The protocol for this study was approved by our Human Studies Committee and permission obtained from the mother of each infant studied. We wish to acknowledge Drs. Jeffrey Perlman and John Roberts for editorial assistance.


  1. Brazelton, T.B. (1973): Neonatal Behavior Assessment Scale, J.B. Lippincott, Philadelphia, PA.
  2. Grimes, D.A. (1978) Routine circumcision of the newborn infant: A reappraisal. Am. J. Obstet. Gynecol., 130, 125-129.
  3. Kaplan, G. W. (1977): Circumcision--an overview. Curr. Prob. Pediatr. 7, 1-33.
  4. Littman, B. and Parmelee, A., Manual for obstetrical complications. Unpublished manuscript available from A. Parmelee at the Department of Pediatrics, Division of Child Development, School of Medicine, University of California, Los Angeles, California, 90024.
  5. Marshall, R. Stratton, W., Moore, J. and Boxerman, S. (1980): Circumcision I: Effects on newborn behavior. Infant Behav. Dev., 3, 1-14.

Address for reprint requests: Richard E. Marshall, M.D., Department of Pediatrics, St. Louis Children's Hospital,
500 S. Kingshighway, P. O. Box 14871, St. Louis, Missouri 63178, U.S.A.

(Revised 08 August 2005)