INTRODUCTION AND PURPOSE
Congenital muscular torticollis (CMT) is a condition characterized by tightness and shortening of the sternocleidomastoid (SCM) muscle on 1 side of the neck.1–3 Infants with CMT display a lateral head tilt to the affected side and rotation to the opposite side that restrict their range of motion in the cervical spine.1–3 Many children with CMT also present with plagiocephaly or craniofacial asymmetry.2,3
The differential diagnosis of CMT to rule out nonmuscular types of torticollis may be achieved by ultrasonography4 and by taking plain film radiographs.1,5 Physical therapists may use several methods to evaluate cervical spine range of motion6,7 and habitual head deviation from midline3 in infants with CMT. Klackenberg et al7 showed high intrarater reliability in measuring passive range of motion in infants with this diagnosis, while using goniometry and a protractor. Another recent study conducted with infants developing typically established reference values for cervical spine range of motion using protractor measurements of lateral flexion and rotation of the neck.8 These reference values may potentially be used to track progress in infants with CMT after establishing baseline measurements.8
The goniometric and protractor measurements of head rotation and lateral flexion assess the amount of asymmetry in passive range of motion of the infant's neck and therefore are measures of impairment. Öhman and Beckung8 suggested measuring neck lateral flexor muscle function in infants developing typically by assessing their lateral head righting in a side-lying horizontal suspension position, with grades from 0 to 4 assigned on an ordinal scale. Although the reliability of this method of measurement for infants with CMT has not been established, the reference values developed by Öhman and Beckung8 potentially may be used when tracking functional progress in patients with this condition.
Still photography used to assess the habitual head deviation from midline in infants with CMT is another functional measure.3 Physical therapy intervention for patients with this condition addresses their functional goal of achieving a midline head position through unrestricted, spontaneous movement in a variety of positions and during movement transitions. Still photography can be used to establish a baseline and to track the child's functional progress. However, the reliability of this method of measurement has not been examined.3 The purpose of this study was to establish intrarater and interrater reliability of still photography used to measure habitual head deviation from midline in infants with CMT as described by Rahlin.3
Participants and Investigators
The study participants were 30 infants (18 boys and 12 girls, chronological ages 4.5-16 months) recruited through 2 Cranial Technologies, Inc, outpatient clinics in Skokie and Oakbrook Terrace, Illinois. These clinics provide cranial molding devices for treatment of deformational plagiocephaly in infants, and many of their patients are also diagnosed with CMT. The investigators were pediatric physical therapists in private practice. Each of them had greater than 15 years of clinical experience. The investigators were not affiliated with each other or with Cranial Technologies, Inc.
As recommended by Portney and Watkins,9 the sample size for this study was estimated by examining previous research7 and by using power and sample size tables.10 In their cervical spine passive range-of-motion reliability study, Klackenberg et al7 had 23 participants, but no still photography reliability studies were available for comparison. To use Pearson r as a method of data analysis, a power and sample size table for correlation10 was used. It was estimated that a sample of 24 participants would be needed to obtain a power of 90% (α2 = 0.05) with at least a moderate correlation coefficient of 0.60. However, a decision was made to recruit at least 30 infants to allow for possible attrition, unforeseen difficulties with picture processing, or other problems.
The participants were included in the study if they had a diagnosis of CMT as documented in their medical records and reported by the clinic therapist. To ensure the validity of measurement, infants diagnosed with other types of torticollis, such as neurological torticollis, Sandifer syndrome, benign paroxysmal torticollis, and other nonmuscular types of torticollis, were excluded from the study. Fourteen participants (9 boys and 5 girls) had left CMT, and 16 participants (9 boys and 7 girls) had right CMT. Six infants (3 girls and 3 boys) were born prematurely. All participants were diagnosed with plagiocephaly; 7 of them had other medical diagnoses, including bilateral hip dysplasia (2 infants), laryngomalacia, cri du chat syndrome, sagittal synosthosis and scaphocephaly, hypotonia, and a congenital heart defect.
Still Photography Procedure
This study was approved by the Institutional Review Board at the Rosalind Franklin University of Medicine and Science, North Chicago, Illinois. The opportunity to participate was offered to the infants' parents by placing flyers in the waiting room and treatment rooms as agreed with the clinic management. In addition, the therapists working with the infants at the clinic discussed the possibility of participation in this research with the patients' parents. When the parents expressed interest in having their child participate in the study, the investigators met with them, explained the details of the project, and asked them to sign the informed consent form. Because of the young age of the study participants, assent to participate was not obtained.
Before initiating data collection, the investigators practiced the still photography procedure to ensure efficiency and accuracy in following all the necessary steps as described later. A doll was used during a practice session to simulate a study participant.
The investigators used a designated clinic room to collect data for the study. The data collection procedure took 5 to 10 minutes per child. One of the investigators asked the parent to undress the child to the diaper and place him or her in a supine position on a floor mat covered with a disposable sheet. The second investigator prepared a card with her initial and the child's participant number and then placed the card on the mat within the field of the photograph she was about to take. The first investigator provided the child with a visual stimulus at midline. No additional steps were taken to place the child's head in a midline position. The second investigator made sure that the camera was set to display the current date and then took a photograph of the child in a supine position using a Canon PowerShot SD 800 IS digital camera. She stepped in, held the camera above the child while looking straight down at him or her, and kept the camera in midline to avoid eliciting neck rotation or hyperextension by the child, essentially replacing the visual stimulus.
Each of the investigators took 2 or more still photographs of the participant positioned supine to obtain 2 photographs acceptable for processing. After each photograph was taken, the investigator reviewed it on the display screen of the camera for optimal results and repeated the same procedure if the participant's gaze diverted from the midline or if the card was not fully visible. The parents were instructed that they could pick up and hold the child after each photograph, if needed for comfort, but did not have to do this if not necessary. They were also instructed that if their son or daughter became uncooperative, the photo session would be restarted once and only if the child could be soothed. The investigators alternated the order in which they took the photographs for each participant, with one of them being first for 1 child and second for the next child.
Photograph Processing and Control for Investigator Bias
At the end of each data collection day, the principal investigator downloaded and printed 2 sets of the 4 photographs taken for each participant on that day. She then placed them in the dated folders designated “B” and “M” (using the investigators' initials) and numbered 1 to 4 for each of the investigators. Folders 1 and 2 contained photographs taken by one investigator, and folders 3 and 4 contained photographs taken by the other investigator. One week after taking the photographs, each investigator obtained her folder 1 and measured each participant's head position as described in the “Measurement Procedure” subsection of this article. The photographs with recorded measurements from folder 1 were placed in an envelope and kept sealed until the data collection for the entire study was completed.
The same procedure was repeated for folder 3 one week later, for folder 2 one week after that, and for folder 4 the following week, so that weekly intervals were maintained between measurements, and each investigator alternated measuring the photographs she took with those taken by the other investigator. Only after the data collection for the entire study was completed, the sealed envelopes were opened, and all measurement data were compiled for analysis. Maintaining weekly intervals between measurements, alternating measurements of the photographs taken by different investigators, and sealing the envelopes guaranteed that the investigators were unable to recall the results of previous measurements and had no access to previously collected data for the same participant when performing the next measurement. This ensured control for investigator bias.
The investigator used a number 2 mechanical pencil with standard lead and a solid plastic ruler to draw 1 line through the participant's eyes and another line through the superior aspect of the acromion processes (at the top of the lateral third of the shoulder) on the printed photograph and then extended the lines until they intersected (Figure 1). Additional blank paper was taped to the edge of the printed photograph if extra space was required to extend the lines. Then, the acute angle between the 2 lines was measured to the nearest degree using a protractor. Both investigators used identical rulers, pencils, and protractors for this procedure. The measurement results were recorded on the paper with the photograph and signed by the investigator. The recorded angle represented the habitual head deviation from the midline exhibited by the infant. A negative value indicated a head tilt to the right (Figure 1A), and a positive value indicated a head tilt to the left (Figure 1B).
After measuring the first set of photographs taken for 8 children recruited on the first day of data collection, the investigators discussed their measurement procedure and discovered that 1 of them had been drawing the line through the child's eyes using the pupils as the target points to connect while the other 1 had been drawing that line by connecting the lateral corners of the child's eyes. Because this might be a potential source of measurement error, the investigators decided to remeasure that set of photographs after the entire data collection was completed and to drop the initially recorded data from the study. The lateral corners of the child's eyes were used to draw the line through the eyes for the rest of the study and for the remeasured pictures.
To determine the intrarater reliability of the still photography, the following measurements were compared for 30 participants: MM1 and MM2—measurements of 2 photographs of each child taken by investigator M; and BB1 and BB2–-measurements of 2 photographs of each child taken by investigator B.
To determine the interrater reliability of the still photography, all possible comparisons were made (n = 12) (see the Appendix). Microsoft Office Excel11 was used to calculate paired t tests to analyze the difference and Pearson r to analyze the agreement in measurements performed by the same rater and by 2 different raters. Using the SPSS base version 16.0,12 intraclass correlation coefficients, ICC (3,1) and ICC (2,1),13 and corresponding 95% confidence intervals were calculated to analyze both, the agreement and difference in measurements taken by the same rater and by 2 different raters, respectively. The variability in the participants' head position was examined using Microsoft Office Excel11 by calculating the measures of central tendency and variability on the basis of the 8 measurements taken for each participant by 2 raters.
Tables 1 and 2 contain the intrarater and interrater reliability estimates, respectively. There was no difference found between measurements taken by the same rater and by 2 different raters (2-tailed P > .05), except for the situation MM1-BM1, with the 2-tailed P < .05 (Table 2). Pearson r varied from 0.80 to 0.85 for the measurements performed by the same rater (Table 1) and from 0.72 to 0.99 for measurements performed by 2 different raters (Table 2), with power estimated at greater than 95% (α2 = 0.05). The ICC (3,1) varied from 0.79 to 0.84 (Table 1), and ICC (2,1) varied from 0.72 to 0.99 (Table 2).
Table 3 contains habitual head-tilt angle values and estimates of central tendency and variability for 8 measurements taken for each participant by 2 raters. The measures of central tendency and variability included means, standard deviations, and the absolute values of mean habitual head-tilt angles for each participant. The absolute value or modulus is defined as “the value of real number disregarding its sign.”14 In this research, the absolute values of the mean head-tilt angles were used to describe the amount of habitual head deviation from midline, regardless of the side of the tilt.
The variability in the head position assumed by each study participant when the still photographs were being taken (Table 3) warranted further examination. To assess this variability in a systematic manner, data were reordered based on the absolute values of mean habitual head-tilt angles and divided into 3 equal groups of 10. When the absolute values of mean habitual head-tilt angles were examined for each of the 3 groups, we discovered that they were below 3°, between 3° and 6°, and greater than 6° in the first, second, and third groups, respectively (Table 4). The average standard deviations were calculated for these 3 sets of absolute values. The results showed that average standard deviation increased as the absolute values of mean habitual head-tilt angles decreased, indicating a greater variability in the participants' head position with smaller habitual head-tilt angles (Table 4).
In the current study, Pearson r and ICC (3,1) varied from 0.80 to 0.85 and from 0.79 to 0.84, respectively (Table 1), which indicated good intrarater reliability.15 For the measurements performed by 2 different raters, both Pearson r and ICC (2,1) varied from 0.72 to 0.99 (Table 2), which indicated moderate to good interrater reliability.15 Clinical interpretation of these results is provided later.
Limitations/Sources of Error and Suggestions for Future Research
When discussing the reliability of measurement, 3 sources of error need to be considered: the rater, the measurement instrument, and the variability of the characteristic that is being measured.15 In this study, 2 raters used a camera to record, and a ruler and a protractor to measure habitual head deviation from midline in infants with CMT. The following possible sources of error need to be considered: (1) possible differences in how the raters followed the standard setup, photograph-taking, and measurement procedures; (2) the extent to which the procedure was standardized; (3) the qualities of the instruments the raters used, including the rulers and the protractors; and (4) the variability in the head position assumed by each study participant when the still photographs were being taken.
The still photography procedure in this study did not specify a set distance between the camera and the participant. The investigators held the camera in their hands and did not use a tripod. We did not feel that greater precision was necessary, because in order to hold the camera directly above the participant, the therapist would need to stand at the child's feet. Therefore, this procedure should be reproducible in the clinic from the description provided in the “Methods” section of this article.
Another potential source of error related to the standardization of the procedure could be the presentation of the visual stimulus because the response to it was potentially dependent on the participant's interest. However, in reality, after the first investigator provided the child with a visual stimulus at midline and the second investigator stepped in with the camera, the vast majority of the children looked directly at the camera, likely out of curiosity. In several instances, when their gaze deviated, the first investigator provided a visual stimulus in midline immediately next to the camera, which proved to be very effective.
As stated earlier, after measuring the photographs taken during the first data collection day, the investigators discovered a difference in how they were drawing the lines through the participants' eyes, which led to a subsequent clarification of the measurement procedure. The statistical analysis of the measurements taken by the 2 investigators revealed no statistical difference between them, except for the situation MM1-BM1, with P < .05 (Table 2). A closer analysis of the standardized measurement procedure revealed that it was not always easy for the investigators to find the acromion processes in each photograph, especially if the child had his or her shoulders flexed or abducted, as in the situation MM1-BM1. To eliminate this problem in the future, we suggest placing small sticker markers on the infant's acromion processes before taking the photograph and using them to draw the line on the printed photograph.
The rulers and the protractors used in this study did not seem to generate a measurement error. However, both investigators observed and agreed that using a clear (see-through) ruler instead of a solid one might be more convenient for drawing accurate lines, and it might help eliminate the need of erasing and redrawing some of the lines if they seemed inaccurate initially.
Finally, the variability in the participants' head position (Table 3) might be the greatest source of error in this research. The study participants assumed a habitual head position spontaneously, without any physical help, except for a visual stimulus the investigator provided for them at midline. This most likely introduced greater measurement variability than would be seen during goniometric or protractor measurements of passive head rotation and lateral flexion. However, while goniometric and protractor measurements are measures of impairment, still photography may be used as a measure of function, because the physical therapy intervention for infants with CMT addresses their functional goal of achieving a midline head position through spontaneous, unrestricted movement in a variety of positions and during transitions between positions.3 Clinically, it has been observed that it is not unusual for children who have been undergoing physical therapy for CMT and displaying improvement in their head and neck alignment to show greater variability in their habitual head position and even occasionally assume a position with a small head tilt to the opposite side.3 The results of the current study confirmed this observation. Furthermore, the examination of the variability in the participants' head position showed that there was greater variability in the head position observed with smaller head-tilt angles and less variability in the head position with larger head-tilt angles as indicated by greater standard deviation values calculated for smaller mean head-tilt measurement values (Table 4, Figure 2). This result was unexpected and needs to be discussed further.
Movement variability has been traditionally viewed as a source of error in motor performance, which would negatively affect the reliability of measurement.16 However, normal movement of a healthy individual is characterized by its inherent optimal variability, which Harbourne and Stergiou16 described as lying “between too much variability and complete repeatability.” This means that such a complex dynamic system as a human organism constantly adapts to changes in its environment and is never completely static. The lack of movement variability negatively affects the ability to perform functional skills.16 To illustrate, a child with CMT who displays significant postural asymmetry in all positions has difficulty bringing his or her head to midline and therefore shows decreased variability in head movement and posture, which, if not treated, becomes a stable behavior (Figure 2A). According to dynamic systems theory, a stable behavior is characterized by small variability of movement.16 If we look at the body of a child with CMT as a complex dynamic system that becomes abnormally static because of the tightness and shortening of the SCM muscle on 1 side of the neck, we can imply that therapeutic intervention for CMT creates a disorganization of the system, leading to increased variability in its behavior (Figure 2B). Thus, as the SCM muscle tightness decreases, the child's ability to move his or her head improves, which in turn increases the variability in head position near midline as a means to explore and adapt to the environment in a new situation. Perhaps, after the exploration of a newly found freedom of movement is completed, the variability of head position near midline should decrease once a new stable behavior of the system is achieved. However, to confirm this point, additional research is needed to investigate the variability in habitual head position in infants developing typically. Furthermore, because of the small sample size in the current study, we would like to emphasize that these findings are only preliminary and need to be taken with caution. Therefore, conducting a significantly larger study of variability in head position in infants with CMT to confirm or refute these results is warranted.
In light of the previous discussion, assuming that the current results are valid, one may look at the increase in variability in head position as an indicator of improved function in infants undergoing therapy for CMT. Therefore, although the variability issue seems to have affected the measures of reliability calculated in this research, it appears appropriate to apply the reliability coefficient interpretation guidelines suggested by Portney and Watkins15 to the study results to confirm that still photography is a reliable measure of habitual head deviation from midline in infants with CMT. We propose that still photography may possibly be used as a means of tracking the child's progress over time in 2 different ways: (1) by looking at the change in the angle of habitual head deviation from midline and (2) by looking at the change in variability in head deviation from midline.
The results of this study suggest that as the child's ability to bring his or her head closer to midline improves, taking and measuring a single photograph of the habitual head position may become less valid for calculating a measurable change achieved by therapeutic intervention. Perhaps, taking 3 or 4 photographs, measuring the head-tilt angles, and obtaining the mean value may be more accurate for making a conclusion about the improvement in the child's head position. To reduce the variability in the head position spontaneously assumed by the child, a change to the setup procedure may be also indicated to include manual positioning of the pelvis in line with the trunk before the visual stimulus is introduced at midline.
Another way to use still photography to track functional progress in a patient with CMT would be plotting the head-tilt angles on a graph after taking 3 or 4 still photographs and calculating the mean angle and its standard deviation, thus documenting a change in variability in the child's habitual head position from one reassessment to the next. Further research will be needed to establish the still photography measurement procedure for tracking patient progress over time and to examine its validity.
This research demonstrates that still photography is a reliable method of measuring functional change in head deviation from midline in infants with CMT. We suggest several modifications to the measurement procedure used in this study, including the following: (1) using the lateral corners of the eyes to draw the line through the eyes; (2) placing small sticker markers on acromion processes before taking the photographs and using these markers for drawing the line through the acromion processes; (3) using a clear ruler to draw the lines; (4) positioning the child's pelvis in line with the trunk before introducing the visual stimulus at midline; and (5) taking 3 or 4 photographs at each reassessment when documenting change over time. These modifications may further improve the reliability of measurement15 and may also make using this method of tracking the child's functional progress in therapy more valid. To test these hypotheses, further research is needed. Future studies should also test the still photography procedure with infants developing typically and compare the variability in habitual head position in these infants and in patients with CMT. To strengthen the results, these studies should use a large sample size.
We would like to thank Karla Grassman, PT; Nancy Haney, PT, MS, PCS; and Sylvia Nielsen, PT, MPH, from Rady Children's Hospital, San Diego, California, for their contribution to the development of the measurement procedure for this research. We are very grateful to the participants and their parents for taking part in this study. We thank management of Cranial Technologies, Inc, for allowing us to recruit research participants from their clinics in Skokie and Oakbrook Terrace, Illinois. We also thank the clinic staff for providing us with a very friendly and supportive environment for this project.
1. Do TT. Congenital muscular torticollis: current concepts and review of treatment. Curr Opin Pediatr. 2006; 18(1):26–29.
2. Hermann MJ. Torticollis in infants and children: common and unusual causes. Instr Course Lect. 2006; 55: 647–653.
3. Rahlin M. TAMO therapy as a major component of physical therapy intervention for an infant with congenital muscular torticollis: a case report. Pediatr Phys Ther. 2005; 17: 209–218.
4. Cheng JC, Metreweli C, Chen TM, Tang S. Correlation of ultrasonographic imaging of congenital muscular torticollis with clinical assessment in infants. Ultrasound Med Biol. 2000; 26: 1237–1241.
5. Snyder EM, Coley BD. Limited value of plain film radiographs in infant torticollis. Pediatrics. 2006; 118:e1779–e1784.
6. Karmel-Ross K, Lepp M. Assessment and treatment of children with congenital muscular torticollis. In: Karmel-Ross K, ed. Torticollis: Differential Diagnosis, Assessment and Treatment, Surgical Management, and Bracing. New York, NY: The Haworth Press Inc; 1997:21–67.
7. Klackenberg EP, Elfving B, Haglund-Akerling Y, Carlberg EB. Intra-rater reliability in measuring range of motion in infants with congenital muscular torticollis. Adv Physiother. 2005; 7: 84–91.
8. Öhman AM, Beckung ERE. Reference values for range of motion and muscle function of the neck in infants. Pediatr Phys Ther. 2008; 20: 53–58.
9. Portney LG, Watkins MP. Statistical inference. In:Portney LG, Watkins MP, eds. Foundations of Clinical Research: Applications to Practice. Upper Saddle River, NJ: Pearson Education Inc; 2009:405–432.
10. Portney LG, Watkins MP. Appendix C: power and sample size. In:Portney LG, Watkins MP, eds. Foundations of Clinical Research: Applications to Practice. Upper Saddle River, NJ: Pearson Education Inc; 2009:830–855.
11. Microsoft Office Excel [computer program]. Part of Microsoft Office Professional Edition 2003. Redmond, WA: Microsoft Corporation; 2003.
12. SPSS [computer program]. Base version 16.0 for Windows. Chicago, IL: SPSS Inc; 2007.
13. Portney LG, Watkins MP. Statistical measures of reliability. In:Portney LG, Watkins MP, eds. Foundations of Clinical Research: Applications to Practice. Upper Saddle River, NJ: Pearson Education Inc; 2009:585–618.
15. Portney LG, Watkins MP. Reliability of measurements. In:Portney LG, Watkins MP, eds. Foundations of Clinical Research: Applications to Practice. Upper Saddle River, NJ: Pearson Education Inc; 2009:77–96.
16. Harbourne RT, Stergiou N. Movement variability and the use of nonlinear tools: principles to guide physical therapist practice. Phys Ther. 2009; 89(3):267–282.
Appendix List of Comparisons Made to Estimate Interrater Reliability Coefficients
1. MM1 and BB1—measurements of the first photograph of each child taken and measured by both investigators.
2. MM2 and BB2—measurements of the second photograph of each child taken and measured by both investigators.
3. MM1 and BB2—measurements of the first photograph of each child taken and measured by investigator M and the second photograph of each child taken and measured by investigator B.
4. MM2 and BB1—measurements of the second photograph of each child taken and measured by investigator M and the first photograph of each child taken and measured by investigator B.
5. MM1 and MB1—measurements of the first photograph of each child taken by each investigator and measured by investigator M.
6. MM2 and MB2—measurements of the second photo-graph of each child taken by each investigator and mea-sured by investigator M.
7. BM1 and BB1—measurements of the first photograph of each child taken by each investigator and measured by investigator B.
8. BM2 and BB2—measurements of the second photo-graph of each child taken by each investigator and measured by investigator B.
9. MM1 and BM1—measurements of the first photographtaken by investigator M and measured by both investigators.
10. MM2 and BM2—measurements of the second photo-graph taken by investigator M and measured by both investigators.
11. BB1 and MB1—measurements of the first photographtaken by investigator B and measured by both investigators.
12. BB2 and MB2—measurements of the second photo-graph taken by investigator B and measured by both investigators.
articular range of motion; infant; neck; neck muscles; photography; reliability; rotation; torticollis/congenital; torticollis© 2010 Lippincott Williams & Wilkins, Inc.