Investigation of Trunk and Extremity Movement Associated With Passive Head Turning in Newborns.

In some approaches to physical therapy, the asymmetrical tonic neck reflex (ATNR) is viewed as a reflex response to turning of the head.[1,2] The response consists of extension of the upper and lower extremities on the side to which the face is turned and flexion of upper and lower extremities on the side to which the skull is turned. The ATNR has been theorized to play an important role in normal development of human movement by facilitating early visual inspection of the hand, and thus eye-hand coordination, as well as by modifying newborn symmetry.[1,2] Similarly, the neonatal righting reflex is another reflex in which the stimulus is head turning. The response is believed to be for the trunk to turn as a block in the direction of the head turn. This righting reflex has been viewed as supporting the infant's ability to align body parts for rolling by causing turning of the body in response to a head turn.[1,2]

Magnus[3] first reported the ATNR and believed that the phenomenon occurs only in pathological conditions in humans. Gesell[4] was the first author to report that the reflex was a "normal and prominent feature of early infant behavior" and asserted that the reflex was present universally at 4 weeks when the infant was in a supine position. Since Gesell's time, numerous investigators[5-10] have reported evidence of the ATNR in newborn infants with no known medical problems. These investigators reported prevalence rates of the reflex among full-term neonates in the range of 40% to 67.5%. Increased prevalence as high as 91% among infants born prematurely and tested at 40 weeks postconceptual age has been reported,[8,11,12] and the ATNR has been reported to be universally present in premature infants over 33 weeks postconceptual age when they are tested at birth.[13]

Very low prevalence rates for ATNR among full-term neonates in the range of 1% to 11% have been reported.[5,11,14-17] These lower rates would suggest to us that the reflex plays a less prominent role, if any, in the development of normal infant movement. Current theories suggest a reduced role played by stimulus-response types of movement patterns in the development of typical human movement.[18] In our view, these lower frequencies, therefore, are more consistent with current theories. The large discrepancies in the frequencies reported, however, have not been explained. Thus, the meaning, if any, of the ATNR in normal motor development is unknown.

One factor that may explain some of the discrepancies that have been reported in prevalence of the ATNR in full-term neonates may be whether the infants were allowed to turn their own heads or whether someone else passively turned their heads. In a study of infants 4 to 10 weeks of age, Coryell et al[19] passively turned the head and used a visual stimulus to elicit active head turning. They found that a full head turn was more likely than a partial head turn to elicit the reflex. Although active and passive head turning were equally likely to elicit the reflex if the head was fully turned, the passive head turn was much more likely to be a full turn than the infant's own active head turn. Several researchers[5,16,17] have reported very low incidences of the ATNR during spontaneous supine postures even when using a visual stimulus to stimulate head turning. Full-term newborn infants typically exhibit a posture of physiologic flexion (passively flexed state of the newborn infant resulting from crowding in the womb and consequent shortening of soft tissues) with the head maintained slightly out of midline when the infant is supine.[20] During studies of spontaneous movement, researchers may not have observed much ATNR posturing because the position of full head turning was only rarely assumed. Forslund and Bjerre,[11] however, found a low rate (8%) of ATNR using an imposed turning of the head, so active versus passive head turning cannot completely explain the discrepancies in previous research findings.

Vassella and Karlsson[10] reported that they observed an ATNR pattern 60% of the time in neonates, but that the number of observations should be adjusted for the fact that the infant may assume the ATNR pattern by random flexion and extension of the limbs at least some of the time. They made the point that, in previous studies of the ATNR, researchers failed to account for random movements of the extremities that may, by chance, mimic the ATNR pattern. In order to adjust for chance, they suggested that only infants assuming the pattern during at least 5 out of 10 head turns should be counted as displaying the ATNR pattern. Using this standard, they reported that only 8% of infants displayed the pattern more often than would occur during random movement of the infant's limbs. The choice of 5 out of 10 head turns is arbitrary, however, and is not based on standard inferential methods. In our study, we used inferential statistics to assess the effect of chance positioning of the limbs by analyzing the association between head and extremity positioning as well as the association between head and trunk positioning using chi-square analysis, following conventional research procedures.

Marinelli[6] has argued that the infant may become habituated to repeated turning of the head (the procedure used in the study of Vassella and Karlsson[10]) and thus fail to assume the ATNR position after the first few turns. In 2 studies that demonstrated a low prevalence of the reflex, the researchers also used a method of repeated passive head turning,[14,15] but Forslund and Bjerre[11] found a low rate (8%) with a single head turn, so habituation cannot completely explain the discrepancies in frequency observed.

Although there is little evidence documenting an association between the ATNR and trunk incurvation (lateral curvature of the trunk), Connolly and Michael[21] have documented a possible relationship between the ATNR and scoliosis in school-age children. In their study, an abnormally strong or questionable ATNR was documented in 6 children with no known medical problems prior to the development of idiopathic scoliosis. The ATNR was rated in the quadruped position by having the child look over the left and right shoulders 4 times. The ATNR was rated abnormally strong if the child had "definite difficulty assuming the quadruped position with flailing of the arms at least 2 of the 4 times, difficulty maintaining the position, or if the child's head movements interrupted balance at least twice."[21] If the child was unable to maintain the extended position of the upper extremity when the face was turned to the opposite shoulder, the ATNR was considered abnormally strong on the side to which the face was turned. The ATNR was rated as questionable if the child "restricted his movements to obtain his balance, if the body posturing appeared rigid or stiff, or if head movements were restricted."[21] Either the convexity of the scoliotic curve matched the side of the predominant ATNR, or the ATNR was bilaterally strong in all 6 children with scoliosis who were available for follow-up. Fiorentino[2] and Bobath[22] have suggested that an abnormal ATNR may be associated with scoliosis in people with neurological involvement, but they provided no data to support their contention. The study of Connolly and Michael and the articles by Fiorentino and Bobath suggest that there may be an association between the ATNR and spinal curvature, despite the fact that no trunk component has been described as part of the ATNR response to head turning.

A trunk reaction in association with head turning in infants was observed in at least 2 studies.[23,24] Weggemann et al[24] reported that a "tilted to one side" position is common in infants. The illustration in their article shows an ATNR posture with the infant's face turned to the left, the lower extremities showing a stronger response than the upper extremities, and strong trunk incurvation, convex to the left.

Schaltenbrand stated that the neck righting reflex "begins with a swing of the pelvis in a direction opposite to the rotation of the head."[23(p723)] This brief description corresponds to the movement observed in our study. Schaltenbrand, however, did not further discuss the frequency or character of the pelvic movement. The primary (newborn, neonatal, neck on body) righting reflex is usually described as the trunk (shoulder and pelvis) turning simultaneously with the head,[1,25,26] in contrast to Schaltenbrand's description of the pelvis moving separately from the shoulders. Because both the primary neck righting reflex and the ATNR are elicited by turning the infant's head, a trunk or extremity response to head turning cannot be identified as being part of either reflex pattern without considering both.

Clopton has evaluated the reflexes, including the ATNR, of numerous newborn infants using the Brazelton protocol[27] and has regularly observed a strong component of trunk incurvation, convex to the face side, in association with the ATNR similar to that described by Schaltenbrand[23] and Weggemann et al.[24] The lateral trunk curvature observed was convex to the face side, and, consequently, the pelvis was laterally tilted. The trunk incurvation response to head turning appeared to the first investigator (NAG) to be stronger and more frequent than the extremity response.

In none of the previous studies of ATNR or the neonatal righting reflex was the position of the infants' head masked when the investigators recorded the presence or absence of extremity or trunk response. Observations of the infants' responses may have been influenced by the observers' expectations that they would or would not see the ATNR posture. Differences in expectations, therefore, might account for some of the discrepancies in the frequency of ATNR reported. Observers' classifications may be distorted to match an expected pattern.[28,29] For that reason, if the investigator expects to see the ATNR posture when the infant's head is turned, he or she may, without intending to, focus on a very brief period when the infant assumes the expected posture and ignore longer periods when the infant does not assume the expected posture. Similarly, if the limbs do not assume the expected posture, the observer's categorization of the limb position may be distorted to resemble the expected ATNR posture. We believe that only if the observer records the extremity position for a specified point in time without knowing the position of the head and records the head position without knowing the position of the extremities can bias due to expectations be reduced.

In our study, we blocked a view of each infant's head position from the investigator who rated extremity and trunk position, and we blocked sight of the extremity and trunk positions from the investigator recording the direction of head turns. In an effort to reduce rater bias and to ensure that a representative sample of the infant's behavior would be rated, we used a standardized time sampling. Observations were made according to a predetermined schedule (ie, every 5 seconds) instead of allowing the observer to select the time periods to observe. We used a single passive head turn because the purpose of our study was to document the presence of the reflex pattern, if possible.

We hypothesized that extremity and trunk postures are associated with head position following a passive head turn. Thus, our hypothesis predicts that flexion of upper and lower extremities and trunk concave incurvation will be observed on the side to which the skull is turned more often than would occur by chance and that extension of the upper and lower extremities and trunk convex incurvation will be observed on the side toward which the face is turned more often than would occur by chance.

Method

Subjects

Fifty newborns between 37 and 42 weeks' gestational age were videotaped between 4 hours and 7 days after birth. All infants were born via uncomplicated spontaneous vaginal delivery or routine cesarean section and weighed at least 2.25 kg (5 lb). None were being treated for any diagnosed medical problem. Parents were contacted in the postpartum unit of University Medical Center in

Lubbock, Tex, and provided informed consent for their infants to participate in the study.

The first 4 subjects were used to determine the best camera angle and placement to record head, trunk, and extremity positions and to standardize procedures. The second 4 infants were used to assess interrater reliability. Data obtained for the remaining 42 subjects were used in the data analysis to test the research hypothesis.

Instrumentation

A video camera mounted on a tripod was placed at the foot end of the newborn's crib to record the infant's movement. The camera tripod center support pole was placed 45.7 cm (18 in) from the crib, with the camera platform 157.5 cm (62 in) from the floor. The camera was set at an angle of 50 degrees from horizontal and focused on the infant. The videotaping was done with 2 teams of 2 physical therapist students each, using one videotape tar each team to record the responses of one half of the infants in the study. The videotapes were then exchanged so that each team analyzed videotapes of infants they had not observed previously.

Procedure

Infants were undressed except for their diapers and videotaped in a supine position. All infants were awake, but not crying, before the test started (Brazelton state 4 or 5).[30] In the pilot study, we found that the infant's whole body tended to turn when we attempted to turn only the head, providing a strong clue as to which way the head was being turned. Because it was essential for the purpose of the study that the investigator rating extremity position be unaware of the infant's head position, we decided to stabilize the infant's shoulders with 2 of the investigator's fingers placed lightly on each shoulder to prevent the infant's whole body from turning when the head was turned. Paine et al[15] also examined the ATNR and stabilized the infant's shoulders in a similar manner, and they did not report any effect on the ATNR response due to stabilizing the infants' shoulders. After the first investigator stabilized the infant's shoulders, a second investigator started videotaping and then moved to the head of the crib and turned the infant's head. The direction of the first head turn was decided by a coin toss. The infant's head was turned in one direction and held for a slow count of 15. The infant's head was returned to midline and allowed to rest in midline briefly, then turned in the other direction and again held for approximately 15 seconds. Previous investigators have found the 15-second time period sufficient to elicit the full ATNR,[11,12,19,23,31] and Pollack[9] found that a 45-second time period elicits too much resistance from the infant.

The videotapes were analyzed by playing and stopping them at each 5-second interval, starting as soon as the infant became visible, on the videotape, so that the infant's position in the video frame at the end of each 5-second period could be recorded. For each interval, one investigator rated the direction of head turn, and a second investigator rated the upper-extremity position, lower-extremity position, and trunk position from the videotape. The investigator who recorded the direction of the head turn stood and placed a cardboard screen perpendicular to the monitor so that the sitting investigator who rated the upper and lower extremities and trunk could not see the direction of the head turn and so that the investigator rating the head turns could not see the extremity or trunk responses. The raters placed their rating sheets so that neither rater could see each other's ratings.

The head turn was recorded as "left," "right," or "neutral" according to the direction of the face. Extremities were rated "left," "right," or "neutral" according to the direction of increased extension. The trunk was rated "left," "right," or "neutral" according to the direction of convexity of trunk incurvation. Responses for upper extremities, lower extremities, and trunk were recorded as "left" or "right" only if there was a clear difference between the 2 sides of the body. If there was not a clear difference, the rating was supposed to be "neutral." The raters focused on the lumbar trunk and pelvic tilt in rating whether trunk convexity was present, based on Schaltenbrand's[23] description and the first author's observations during administration of the Brazelton Neonatal Behavioral Assessment to infants as described.

Gesell[4] suggested that there is a delay of 1 to 6 seconds and Marinelli[6] suggested that there is a delay of 1 to 7 seconds in most full-term human infants after the head is turned before the ATNR begins. For that reason, the first 5-second frame after the infant became visible on the videotape and the first 5-second frame following head turns were omitted from the data analysis (range = 2-5 frames per infant, [bar]X = 4.07 omitted from analysis).

All raters in our study were trained to interrater agreement by the first author, using videotapes of the second 4 infants entered into the study. Interrater agreement was computed by determining the number of times the 2 raters agreed in rating the response as "left," "right," or "neutral" divided by the total number of frames rated. Interrater agreement was at least 83.7% for extremities and trunk and 95.7% for the direction of head turn.

An average of 6.71 frames per infant were included in the analyses (range = 3-16 frames per infant). The position of both upper extremities could not be seen on the videotape in 7 frames, and the position of both lower extremities could not be seen on the videotape in 5 frames. The trunk was not visible in 3 frames. Those data were excluded from the analysis. The head was visible in all 282 frames analyzed. The head was turned right in 102 of the frames analyzed and left in 91 frames and was in the neutral position in 89 frames.

Results

The ratings for the 42 infants in the study are summarized in the Table and the Figure. Chi-square analysis revealed a clear association between head position and position of the upper extremities ([chi square = 10.90, df = 4, P [is less than] .03), lower extremities ([Chi square] = 15.42, df = 4, P [is less than] .004), and trunk ([Chi square] = 83.15, df = 4, P [is less than] .001). To remove the effect of predominant infant symmetry, the chi-square tests were partitioned using the procedure of Bresnahan and Shapiro[32] to eliminate the neutral category from analysis for the head, upper extremities, lower extremities, and trunk, thus, comparing only the responses to the right or left side. The strong association between head position and position of the upper extremities ([Chi square] = 6.74, df = 1, P [is less than] .01), lower extremities ([Chi square] = 13.27, df = 1, P [is less than] .01), and trunk ([Chi square] = 60.64, df = 1, P [is less than] .01) remained when only the right and left responses were retained.

Table.

Frequency Counts of Head Turns Left, Right, or Center; Trunk Incurvation Convex Right, Left, or Symmetrical; and Upper- and Lower-Extremity Position in Extension Greater or Less Than the Contralateral Side or Equal to the Contralateral Side(a)

                           Head     Head     Head
                           Left    Center    Right

Upper extremity
  Left                      25       18       17
  Neutral                   58       47       41
  Right                     14       24       31
  Not visible                5        0        2

Lower extremity
  Left                      35       19       17
  Neutral                   55       57       48
  Right                     10       12       24
  Not visible                2        1        2

Trunk
  Left                      69       25       16
  Neutral                   32       59       46
  Right                      1        5       26
  Not visible                0        0        3

(a) "Left" indicates more extension for the left extremity, incurvation convex left for the trunk, or face turned left for the head. "Right" indicates more extension for the right extremity, incurvation convex right for the trunk, or face turned right for the head. "Neutral" indicates bilateral extremities with equal extension and flexion, trunk symmetrical, or head midline.

[Figure ILLUSTRATION OMITTED]

The direction of trunk incurvation, when present, was more likely to agree with face position to the left (68%) than to the right (30%). With the head neutral, there was also a tendency for the trunk to show a curve to the left (83% of the time when a curve was present), although trunk symmetry was the predominant position when the head was neutral (66% of time). If only the matching responses to the left and right are considered (eg, extension of the right extremity or trunk incurvation convex to the right when the head is turned right) and the trunk is compared with the upper and lower extremities, there is a difference ([Chi square]=11.80, df=2, P [is less than] .003), indicating that the trunk shows more responsiveness to head turning than the extremities do. Partitioning the chi-square tests so that the trunk is compared with the pooled responses of the upper and lower extremities indicated that the trunk response is more likely to be asymmetrical than that of the extremities ([Chi square]=9.81, df=1, P [is less than].01), with trunk incurvation to the left more frequent than to the right. The trunk response to the right is less than the extremity response to the right, whereas the trunk response to the left is greater than that of the extremity response to the left.

Of the 183 frames analyzed where both upper and lower extremities were visible and the head was turned right or left, the infant exhibited the ATNR pattern in both upper and lower extremities in only 9 frames (4.92%). The infant's head was turned to the right in 6 of these frames (5.88%) and to the left in 3 frames (3.30%). Of those 9 frames, the trunk was also curved in the expected direction 4 times (1.42%), 2 to the right (1.96%) and 2 to the left (2.20%).

Discussion

Our study provides evidence that head position influences the position of the extremities and trunk in neonates with no known medical problems when a head turn is imposed on the infant. Documenting that head turning exists is, in our opinion, the first step toward defining the importance, if any, of these reflex patterns for normal human development.

Statistical significance, however, does not automatically indicate clinical importance. The full ATNR posture involving both upper and lower extremities is observed only about 4.92% of the time in response to an imposed head turn. This finding suggests that observers reporting that the full response is frequently observed in neonates may have been inadvertently influenced by their expectations that they would see the classic ATNR pattern when judging head and extremity position simultaneously.

The implications of our findings for the developing motor performance abilities of children developing normally or children with neurological deficits is less clear. We used a passive head turn because previous research indicated that that procedure should provide the best possibility of observing an ATNR response.[19] Our decision to passively turn the head is controversial because current evaluation tools for infant movement emphasize observation of spontaneous infant movement as a more consistent method of evaluating infant movement than the elicitation of response to passive manipulations.[18,33] Although eliciting reflexes by passive handling remains a method of detecting the effects of neurological damage in infancy,[34,35] the abnormal stereotyped responses observed in infants who have neurological damage may bear little resemblance to the highly variable responses seen in normal infants without neurological damage.[36] Because a passive head turn was imposed on the infants, inferences about the effects of an active head turn cannot be made.

The infants' shoulders were stabilized to allow unbiased ratings of the infants' responses. Stabilization of the shoulders may have strengthened the infants' trunk responses[1] or may have modified the trunk or extremity responses in other unpredictable ways.

The asymmetrical nature of the trunk response, which was much stronger with the face to the left than to the right, was an unexpected finding. Pollack[9] reported that the extremity ATNR response was stronger to the left, but we did not find a difference in the responses of the extremities. The trunk was observed to curve convex to the left more frequently than to the right even when the head was neutral, providing no stimulus for ATNR or neonatal righting reflex. Asymmetry was much more pronounced, however, when the head was turned. Dunn[37] reported that fetuses tend to lie with their spines to the mother's left side about twice as often as to the mother's right side. Transient infantile idiopathic scoliosis is preponderantly left-sided (76%),[38] suggesting that intrauterine positioning may favor trunk convexity to the left. The asymmetry observed in the trunk response may be related to intrauterine positioning.

Our study raises the question of why there is a traditional division between the ATNR and neonatal righting reflex when both are thought to be elicited by turning the head. Perhaps the current rethinking of the concepts of the stimulus-response model of reflexes to take into account the great variety of factors other than simple neurological maturation that can modify reflexes[18] should include consideration of the validity of the traditional identification of reflex patterns.

Conclusions

Extremity and trunk responses to passive head turning can be documented in newborn infants. The complete ATNR response of flexion in the upper and lower extremities on the skull side and extension of the upper and lower extremities on the face side, however, occurs only about 4.92% of the time. The response to passive head turning includes a component of trunk incurvation, convex to the face side. The trunk response is observed more frequently with the face to the left than to the right and is more frequently observed than either the upper -- or lower-extremity response.

References

[1] Barnes MR, Crutchfield CA, Heriza CB. Reflexes in Motor Development. Atlanta, Ga: Stokesville Publishing Co; 1978:79-88. The Neurophysiological Basis of Patient Treatment; vol 2.

[2] Fiorentino MR. A Basis for Sensorimotor Development, Normal and Abnormal: The Influence of Primitive, Postural Reflexes on the Development and Distribution of Tone. Springfield, Ill: Charles C Thomas, Publisher; 1981:78-83.

[3] Magnus R. Body Posture: Experimental-Physiological Investigations of the Reflexes Involved in Body Posture, Their Cooperation and Disturbances. Rosanoff WR, trans. Berlin, Germany: Verlag Von Julius Springer; 1924:53-255.

[4] Gesell A. The tonic neck reflex in the human infant: morphogenic and clinical significance. J Pediatr. 1938;13:455-464.

[5] Coryell J, Cardinali N. The asymmetrical tonic neck reflex in normal full-term infants. Phys Ther. 1979;59:747-753.

[6] Marinelli PV. The asymmetric tonic neck reflex: its presence and significance in the newborn. Clin Pediatr (Phila). 1983;22:544-546.

[7] Mehlman J. The tonic neck reflex in newborn infants. J Pediatr. 1940;16:767.

[8] Pacella BL, Barrera SE. Postural reflexes and grasp phenomena in infants. J Neurophysiol. 1940;3:213-218.

[9] Pollack SL. The grasp response in the neonate: its characteristics and interaction with the tonic neck reflex. Arch Neurol. 1960;5:574-581.

[10] Vassella F, Karlsson B. Asymmetric tonic neck reflex: a review of the literature and a study of its presence in the neonatal period. Dev Med Child Neurol. 1962;4:363-369.

[11] Forslund M, Bjerre I. Neurological assessment of preterm infants at term conceptional age in comparison with normal full-term infants. Early Hum Dev. 1983;8:195-208.

[12] Konishi Y, Cioni G, Prechtl HF. Posture in preterm infants is not a good indicator of brain damage [English abstract]. Brain Dev. 1993; 25:8.

[13] Allen MC, Capute AJ. The evolution of primitive reflexes in extremely premature infants. Pediatr Res. 1986;20:1284-1289.

[14] Bartlett D, Piper M, Okun N, et al. Primitive reflexes and the determination of fetal presentation at birth. Early Hum Dev. 1997;48: 261-273.

[15] Paine RS, Brazelton TB, Donovan DE, et al. Evolution of postural reflexes in normal infants and in the presence of chronic brain syndromes. Neurology. 1964;4:1036-1048.

[16] van Kranen-Mastenbroek VHJM, Folmer KB, Caberg HB, et al. The influence of head position and head position change on spontaneous body posture and motility in full-term AGA and SGA newborn infants. Brain Dev. 1997;19:104-110.

[17] Vles JSH, van Oostenbrugge R, Kingma H, et al. Influence of head position-change on body posture in pre-term infants (ATNR). Neuropediatrics. 1988;19:96-100.

[18] Piper MC, Darrah J. Motor Assessment of the Developing Infant. Philadelphia, Pa: WB Saunders Co; 1994:6-13.

[19] Coryell J, Henderson A, Liederman J. Factors influencing the asymmetrical tonic neck reflex in normal infants. Physical and Occupational Therapy in Pediatrics. 1982;2:51-65.

[20] Bly L. Motor Skills Acquisition in the First Year. Tucson, Ariz: Therapy Skill Builders; 1994.

[21] Connolly BH, Michael BT. Early detection of scoliosis: a neurological approach using the asymmetrical tonic neck reflex. Phys Ther. 1984;64:304-307.

[22] Bobath B. A study of abnormal postural reflex activity in patients with lesions of the central nervous system: I. Physiotherapy. 1954;40: 259 -280.

[23] Schaltenbrand G. The development of human motility and motor disturbances. Arch Neurol Psychiatry. 1928;20:720-730.

[24] Weggemann T, Brown JK, Fulford GE, Minns RA. A study of normal baby movements. Child Care Health Dev. 1987;13:41-58.

[25] Capute AJ, Accardo PJ, Vining EPG, et al. Primitive Reflex Profile. Baltimore, Md: University Park Press; 1978:36-39. Monographs in Developmental Pediatrics; vol 1.

[26] Fiorentino MR. Reflex Testing Methods for Evaluating CNS Development. 2nd ed. Springfield Ill: Charles C Thomas, Publisher; 1973:14.

[27] Clopton NA, Martin AS. A criticism of reliability procedures on the Brazelton Neonatal Behavioral Assessment. Physical and Occupational Therapy in Pediatrics. 1985;4:55-65.

[28] Goldstein M, Goldstein IF. How We Know: An Exploration of the Scientific Process. New York, NY: Plenum Press; 1978:12-17.

[29] Mitchell M, Jolley J. Research Design Explained. 2nd ed. Fort Worth, Tex: Harcourt Brace Jovanovich College Publishers; 1992:106-109.

[30] Brazelton TB, Nugent JK. Neonatal Behavioral Assessment Scale. 3rd ed. London, England: Cambridge University Press; 1995:31. Clinics in Developmental Medicine, No. 137.

[31] Zafeiriou DI, Tsikoulas IG, Kremenopoulos GM. Prospective follow-up of primitive reflex profiles in high-risk infants: clues to an early diagnosis of cerebral palsy. Pediatr Neurol. 1995;13:148-152.

[32] Bresnahan JL, Shapiro MM. A general equation and technique for the exact partitioning of chi-square contingency tables. Psychol Bull. 1966;66:252-262.

[33] Miller LJ, Roid GH. The TIME: Toddler and Infant Motor Evaluation. Tucson, Ariz: Therapy Skill Builders; 1994:20-22.

[34] Nickel RE. The Manual for the Infant Motor Screen. Eugene, Ore: Child Development and Rehabilitation Center, The Oregon Health Sciences University; 1987.

[35] Wolf M-J, Wolf B, Beunen G. Validity of the infant motor screen in African infants with a low Apgar score. Pediatric Physical Therapy. 1999;11:67-72.

[36] Touwen BCL. Variability and stereotypy in normal and deviant development. In: Apley J, ed. Care of the Handicapped Child. Philadelphia, Pa: JB Lippincott Co; 1978:99-110. Clinics in Developmental Medicine, No. 67.

[37] Dunn PM. Perinatal observations on the etiology of congenital dislocation of the hip. Clin Orthop. 1976;119:11-22.

[38] Wynne-Davies R. Infantile idiopathic scoliosis: causative factors, particularly in the first six months of life. J Bone Joint Surg Br. 1975;57:138-141.

NA Clopton, PT, PhD, is Associate Professor, Physical Therapy Program, Texas Tech University Health Sciences Center, 3601 4th St, Lubbock, TX 79430 (USA) ([e-mail]). Address all correspondence to Dr Clopton.

T Duvall, PT, B Ellis, PT, M Musser, PT, and S Varghese, PT, are recent graduates of the Physical Therapy Program, Texas Tech University Health Sciences Center.

Dr Clopton, Mr Ellis, and Ms Musser provided concept and research design, writing, and data collection and analysis. Dr Clopton provided consultation, and Ms Duvall, Mr Ellis, Ms Musser, and Ms Varghese contributed to planning, data collection and analysis, fund procurement, provision of subjects and facilities/equipment, and writing. Ms Musser provided consultation and clerical assistance. Dr James R Clopton and Dr Steve F Sawyer assisted with statistical analysis and consultation and advice. Dr Clayton Gable assisted with data analysis. Tim Howell assisted with provision of subjects and institutional liaisons to the newborn nursery at the University Medical Center, Lubbock, Tex. The physical therapy class of 1998 helped raise funds for the study.

This study was approved by the Institutional Review Board for the Protection of Human Subjects of Texas Tech University Health Sciences Center and University Medical Center, Lubbock, Tex.

This article was submitted February 25, 1999, and was accepted September 29, 1999.