- Email: [email protected]
Intraexaminer and interexaminer reliability for palpation of the cranial rhythmic impulse at the head and sacrum
Journal of Manipulative and Physiological Therapeutics Volume 24 • Number 3 • March/April 2001 0161-4754/2001/$35.00 + 0 76/1/113773 © 2001 JMPT
Intraexaminer and Interexaminer Reliability for Palpation of the Cranial Rhythmic Impulse at the Head and Sacrum Robert W. Moran,a and Peter Gibbons, MB, BS, DO, DM-SMedb
ABSTRACT Background: A range of health care practitioners use cranial techniques. Palpation of a cranial rhythmic impulse (CRI) is a fundamental clinical skill used in diagnosis and treatment with these techniques. There has been little research establishing the reliability of CRI rate palpation. Objective: This study aimed to establish the intraexaminer and interexaminer reliability of CRI rate palpation and to investigate the “core-link” hypothesis of craniosacral interaction that is used to explain simultaneous motion at the cranium and sacrum. Design: Within-subjects, repeated-measures design. Subjects: Two registered osteopaths, both with postgraduate training in diagnosis and treatment, using cranial techniques, palpated 11 normal healthy subjects. Methods: Examiners simultaneously palpated for the CRI at the head and the sacrum of each subject. Examiners indicated the “full flexion” phase of the CRI by activating silent foot
INTRODUCTION Various forms of cranial manipulation are practiced by a variety of practitioners, including chiropractors, osteopaths, physical therapists, and massage therapists. These techniques have been variously referred to as osteopathy in the cranial field (OCF), cranial osteopathy, craniosacral therapy, sacrooccipital technique, and somatic cranial work. Of these, osteopathy in the cranial field, as first described by W.G. Sutherland1 in 1939, and craniosacral therapy, as popularized by Upledger2 since the late 1970s, are probably the most commonly practiced. For the purposes of this paper, OCF and craniosacral therapy will be considered together because both rely on the concept of involuntary cranial and sacral bone motion and on palpation of a rhythm or impulse in diagnosis a Lecturer, School of Health Sciences, Victoria University, Melbourne, Australia. b Associate Professor, School of Health Sciences, Faculty of Human Development, City Campus, Victoria University, Melbourne, Australia. Submit reprint requests to: Peter Gibbons, MB, BS, DO, DMSMed, School of Health Sciences, Faculty of Human Development, City Campus, Victoria University, PO Box 14428 MC, Melbourne 8001, Australia. E-mail: [email protected]. Paper submitted January 19, 2000; in revised form March 24, 2000.
doi:10.1067/mmt.2001.113773
switches that were interfaced with a computer. Subject arousal was monitored using heart rate. Examiners were blind to each other’s results and could not communicate during data collection. Results: Reliability was estimated from calculation of intraclass correlation coefficients (2,1). Intrarater reliability for examiners at either the head or the sacrum was fair to good, significant intraclass correlation coefficients ranging from +0.52 to +0.73. Interexaminer reliability for simultaneous palpation at the head and the sacrum was poor to nonexistent, ICCs ranging from –0.09 to +0.31. There were significant differences between rates of CRI palpated simultaneously at the head and the sacrum. Conclusions: The results fail to support the construct validity of the “core-link” hypothesis as it is traditionally held by proponents of craniosacral therapy and osteopathy in the cranial field. (J Manipulative Physiol Ther 2001;24:183-90) Key Indexing Terms: Palpation; Reliability; Osteopathic Manipulation; Osteopathy
and treatment of dysfunction. Likewise, the terms primary respiratory mechanism,1 craniosacral motion/rhythm,1 primary respiratory impulse,2 Sutherland wave,3 the Tide,4 and cranial rhythmic impulse (CRI),5 will be used synonymously. The existence of the CRI, as described by Sutherland1 and others, is yet to be definitively demonstrated through use of modern instrumentation. Because there is currently no objective criterion standard or instrumentation of known validity that can be used to measure the CRI, subjective palpation for CRI characteristics persists as the only diagnostic method used in clinically evaluating this rhythm. This lack of objective measurement makes the issue of reliability of palpation as a diagnostic tool important both to clinicians in private practice and to those who teach these techniques. Because of this, great emphasis is placed in teaching skills for CRI palpation to beginners in both postgraduate and undergraduate cranial technique courses. Despite the apparent importance of the role of palpation in diagnosis and treatment, there has been little published research that examines reliability or validity for palpation of CRI rate or indeed of any other clinical characteristic, such as the site or nature of the perceived dysfunction. Likewise, the lack of a criterion standard for CRI measurement means that the validity of palpation as a measure of the CRI has not been established.
183
184
Journal of Manipulative and Physiological Therapeutics Volume 24 • Number 3 • March/April 2001 Cranial Rhythmic Impulse • Moran and Gibbons
Importance of Reliability Using diagnostic methods of known reliability is important for accurate clinical decision-making.6-8 Both Bogduk8 and Leboeuf-Yde9 make persuasive arguments for establishing the reliability of diagnostic measures and implementing these tests in clinical practice. Their arguments are summarized in the following points: 1. In an educational environment, the time and money spent on teaching and learning techniques that remain unvalidated might not be justifiable. 2. In clinical practice, the application of many therapeutic techniques is based on the results of diagnostic procedures; these procedures should be validated and reliable. 3. Subsequent therapeutic decisions can be confounded by inappropriate patient allocation for different interventions, thereby compromising patient outcomes. 4. The notion of preventive care in asymptomatic patients might be unjustifiable if such care is based on unvalidated and unreliable diagnostic procedures.
Concepts of Craniosacral Interaction One of the central concepts in both OCF and craniosacral therapy is that of rhythmic involuntary motion of the sacrum related to the involuntary motion of the occiput. Sutherland was the first to propose such a link between the occiput and sacrum.4 Magoun3 named this the “core-link” and described an involuntary movement of the sacrum caused by a lifting force exerted on the sacrum by the attachments of the spinal dura to the sacrum. Magoun proposed that the effect of this was to rotate the sacrum around a transverse axis at the level of S2. The lifting force is apparently provided by the anterosuperior movement of the foramen magnum during the flexion phase of the CRI. Lay10 reports that the involuntary motion of the sacrum is synchronous with the involuntary motion of the occiput. The concept of the link between occiput and sacrum is widely recognized by proponents of both craniosacral therapy and OCF.2,10-12 Despite the apparent importance of this concept, there has been only one published experimental study that provides any supportive evidence for its existence. Zanakis et al13 recorded movement of surface markers on skin overlying cranial bones while an experienced practitioner simultaneously palpated apparent sacral motion and indicated “flexion” using a foot switch. Zanakis et al reported a 92% level of percentage agreement and concluded that their findings provided support for a hypothesized synchronous craniosacral relationship. Two recent studies have presented data that call the “corelink” concept into question. Rogers et al14 and Norton15 conducted palpation studies in which two examiners were able to palpate for the CRI in the same subject simultaneously from different positions. Norton’s examiners were positioned at the head and sacrum, whereas in the study of Rogers et al the examiners were at the head and feet. Both studies report low levels of agreement between simultaneous findings at the two positions. Chaitow11,16 presents an argument against the linking hypothesis. Chaitow cites the
clinical observations of Butler,17 who reports that to clinically identify adverse neural tension within the neuroaxis, slack must be taken up by placing the patient in a “slump position” of full neck and trunk flexion. Chaitow argues that in the light of such observation, “the core-link hypothesis is probably seriously flawed.”11
Previous Reliability Studies for CRI Rate Upledger18 investigated the reliability of pairs of examiners performing craniosacral examinations on 25 preschool children. Although CRI rate was one of the primary measures of the examination, the results were not published. Subsequent statistical analysis of Upledger’s raw CRI rate data by Wirth-Pattullo and Hayes19 resulted in the calculation of an intraclass correlation coefficient (ICC) of +0.57. However, Upledger’s report fails to identify the method by which CRI rate was measured. This omission makes it impossible to adequately critique Upledger’s conclusions that acceptable interexaminer reliability was attained and that this lends “considerable evidence to the existence of a real and perceptible craniosacral motion system.”18 On the basis of the method as presented, we have difficulty reconciling Upledger’s results with his conclusions. Wirth-Pattullo and Hayes19 conducted an interrater reliability study with 3 physical therapists trained in craniosacral therapy. Each examiner independently palpated the heads of 12 children and adults. Heart and respiratory rates of both examiners and subjects were manually recorded. Correlations between craniosacral rate and subject and examiner heart and respiratory rates were low and not statistically significant. Interrater reliability of craniosacral rate was estimated from calculation of the ICC.2,1 The authors conclude that the resulting reliability coefficient of –0.02 was unacceptable for clinical decision-making. Hanten et al20 sought to determine interexaminer and intraexaminer reliability for palpation of the CRI at the heads of 40 healthy adults. Hanten et al also recorded heart and respiratory rates of examiners and subjects. Two examiners, each with 9 months of experience in palpating craniosacral rhythm, achieved excellent intraexaminer reliability coefficients (ICC [1,1]) of +0.78 and +0.83. However, the interexaminer reliability coefficient of +0.22 was poor. Interrater and intrarater reliability for the measurement of CRI rate was also investigated by Rogers et al.14 They used a repeated-measures design whereby each of 2 trained examiners palpated for the CRI simultaneously in the same subject (n = 28). Examiners activated foot switches in time with palpated events to enable calculation of a CRI rate. One examiner would palpate at the head while the other examiner palpated at the feet. The authors report interrater reliability coefficients (ICC [2,1]) of +0.08 at the head and +0.19 at the feet. Intrarater ICCs range from +0.18 to +0.30, and craniosacral rates recorded simultaneously at the head and feet are reported as being different. Rogers et al conclude that their results suggest that examiners were measuring different phenomena and that they may have been attempting to measure something that did not exist.
Journal of Manipulative and Physiological Therapeutics Volume 24 • Number 3 • March/April 2001 Cranial Rhythmic Impulse • Moran and Gibbons
Objectives and Purposes The purposes of this study were as follows: 1. To examine the interexaminer and intraexaminer reliability of measurement of CRI rate by palpation 2. To examine the relationship between CRI measurements recorded simultaneously at the head and sacrum by 2 examiners (we examined construct validity of the “corelink” theory that the CRI at the sacrum should be the same as that at the head, and we sought to test the null hypothesis that there should be no difference in CRI rates recorded by 2 examiners palpating independently but simultaneously at the head and sacrum) 3. To document the rate of the CRI as palpated by 2 examiners in normal healthy subjects (n = 11) and compare these findings with those reported in the literature.
METHODS Participants A convenience sample of 11 healthy volunteers was recruited by the investigators through word of mouth. Five subjects were male and 6 female. The mean age was 25.8 (SD = 8.0) years and the range 18-44 years. To be eligible to participate, each subject had to be between 18 and 80 years of age and able to lie quietly supine on a treatment table for 45 minutes and could not currently be undergoing any manual therapy or medical treatment. At the time of the study, none of the subjects were experiencing any pain or discomfort that would lead them to seek health care advice or treatment. The study was approved by the Victoria University Human Research Ethics Committee. The procedure was explained to the subjects before it was begun, and they were free to withdraw from the study at any stage. All subjects read and signed informed consent forms. All subjects completed the study.
procedures at the alternate positions. Examiners at the head and sacrum were separated by a large, triple-thickness curtain that hung across the subject at the level of the chest. This design allowed for the collection of 2 independent recordings, taken simultaneously at 2 different positions on the same subject (eg, with examiner A palpating at the head, examiner B simultaneously palpated at the sacrum). Heart rate was recorded in an attempt to monitor the state of arousal of the subject from trial to trial. The procedure was designed to allow for analysis of interexaminer and intraexaminer reliability.
Variables The dependent variable in the study was rate of the CRI, the units being cycles.min–1; the independent variables were examiner (A or B), position (head or sacrum), and measurement trial (1 or 2). For each of the 11 subjects, 8 CRI rate measurements were calculated, as follows: examiner A at the head, trial 1; examiner B at the sacrum, trial 1; examiner A at the head, trial 2; examiner B at the sacrum, trial 2; examiner A at the sacrum, trial 1, examiner B at the head, trial 1; examiner A at the sacrum, trial 2, and examiner B at the head, trial 2.
Instrumentation
Two Victorian State–registered osteopaths participated in the study as examiners. At the time of the study, examiner A had been in private practice for a period of 4.5 years and had completed 2 Sutherland Cranial Teaching Foundation (SCTF) basic-level courses. Examiner A estimated that 100% of patients under their care would undergo at least some cranial treatment and that approximately 30% to 40% of patients would undergo cranial treatment as the major treatment modality. Examiner B had been in private practice for a period of 6.5 years and had completed 2 SCTF basiclevel courses and 1 SCTF intermediate-level course. Examiner B estimated that nearly 100% of patients under their care would undergo at least some cranial treatment and that nearly 100% of patients would receive cranial treatment as the major treatment modality.
Two momentary-action foot switches (model 316-901, RS Components Pty Ltd, VIC, Australia) were activated in time with the designated “full flexion” phase of the CRI as palpated by the examiners. To blind examiners to any auditory cues, the foot switches were positioned inside soundproof housings and a computer fan provided a source of white noise in the background. The operation of the foot switches in the housings was inaudible. Sufficient time was allocated for the examiners to practice activation of the foot switches and become comfortable with their use. Evaluation of correct operation of the foot switches was undertaken before the commencement of formal data collection. Examiners activated their foot switches in response to a visual stimulus presented simultaneously to both examiners. In 3 trials, the 2 examiners were 100% reliable in responding to the visual stimuli and demonstrated perfect interexaminer agreement (for trial 1, ICC [2,1] = 1.0, F5,5 = 5835.84 , and P < .001; for trial 2, ICC [2,1] = 1.0, F8,8 = 9711.36,and P < .001; for trial 3: ICC [2,1] = 1.0, F6,6 = 13574.57, and P < .001). Each subject wore a standard heart rate monitor chest strap (Polar X-trainer Plus, Polar Electro Oy, Finland) throughout the study. Heart rate data were relayed via telemetry to a transceiver positioned near the subject. A custom-designed computer software program (LabVIEW version 5.01, National Instruments, Austin, Tex) was used to record subject heart rate and examiner foot switch data. Data were captured by means of an analog-to-digital board connected to an IBM-compatible personal computer.
Outline of Study Design
Procedures
The study used a within-subjects, repeated-measures design, as previously outlined by Rogers et al.14 Each examiner palpated each subject at the head or the sacrum for 2 trials before the examiners exchanged positions and repeated the
Before the study, the 2 examiners met with the investigator to define the event that was to be recorded by the activation of the foot switch. After discussion, the examiners agreed that they would activate the switch at the
Examiners
185
186
Journal of Manipulative and Physiological Therapeutics Volume 24 • Number 3 • March/April 2001 Cranial Rhythmic Impulse • Moran and Gibbons
Fig 1. Simultaneous CRI rates for examiner A at head and examiner
Fig 3. CRI rates for both examiners at head. Each value is mean ± SE.
B at sacrum. Each value is mean ± SE.
Fig 2. Simultaneous CRI rates for examiner B at head and examiner A at sacrum. Each value is mean ± SE.
instant they recognized the palpatory sensation of the CRI phase described as “full flexion.” Examiners were free to nominate the palpation contact of their own choosing and to vary this contact between subjects according to their own judgment. The subject was positioned supine on a standard treatment table with the heart rate monitor chest strap attached. The curtain was positioned across the subject and secured. The subject rested quietly for 5 to 7 minutes to allow the heart rate to become stable before commencement of data collection. After this period, the examiners were instructed to take up their contacts and were allowed sufficient time to establish a strong sense of the CRI before any data were collected. When the examiners expressed confidence that they had a sense of the CRI, they independently indicated to the investigator that they were ready to commence formal data collection. When both examiners were ready, a 2minute trial began with an instruction to the examiners to “Start data collection.” During this period, the examiners activated their respective foot switches at the moment they sensed the “full flexion” phase of the CRI. After 2 minutes, the investigator announced “Stop palpation” and the examiners released their contacts and relaxed for a minimum of 2 minutes. After the initial trial, there was a second 2-minute
Fig 4. CRI rates for both examiners at sacrum. Each value is mean (± SE). trial with the examiners positioned in the same locations. After 2 trials, the examiners exchanged positions before commencing 2 more trials in the alternate positions. Each examiner thus undertook four 2-minute trials (2 at the head and 2 at the sacrum) per subject. Each trial was of 2 minutes’ duration, and there was a minimum rest period of 2 minutes between trials. Examiner starting positions were alternated between consecutive subjects to avoid any systematic error. Data were collected over a 6-hour period. The examiners had 5-minute rest periods between subjects, and neither examiner felt unduly fatigued by the procedure.
Data Reduction and Statistical Analysis Raw data from the CRI data acquisition software were exported to Microsoft Excel (Microsoft, Redmond, Wash) for descriptive analysis. CRI rates were calculated from the interval times between events recorded from the activation of each examiner’s foot switch. This method resulted in the calculation of instantaneous CRI rates for each trial. The mean CRI rate for each trial was calculated from the instantaneous rates. Mean CRI rates for each subject were calculated from the pooled trial data in each position.
Journal of Manipulative and Physiological Therapeutics Volume 24 • Number 3 • March/April 2001 Cranial Rhythmic Impulse • Moran and Gibbons
Table 1. Results for palpation of CRI rate at head and sacrum for both examiners ICC (2,1) coefficient Intraexaminer Examiner A at head Examiner A at sacrum Examiner B at head Examiner B at sacrum Examiner A at head and sacrum Examiner B at head and sacrum Interexaminer Examiner A at head, examiner B at sacrum Examiner B at head, examiner A at sacrum Both examiners at sacrum Both examiners at head
F value*
Pearson r coefficient
P value
+0.47 +0.65 +0.73 +0.52 –1.46 –1.47
2.61 5.17‡ 8.35‡ 4.71‡ –0.03§ –0.07‡
+0.47 +0.68 +0.80 +0.74 +0.31 +0.62
0.144 0.022‡ 0.003‡ 0.009‡ 0.164 0.002‡
–0.09 +0.31 –0.02 +0.05
0.54 2.18 0.93 1.27
–0.314 +0.45 –0.035 +0.17
0.320 0.158 0.920 0.640
CRI mean rate ± 95% CI (cycles.min–1)
CRI mean rate (SD) absolute difference (cycles.min–1)
Difference in mean CRI rate P value†
2.92 ± 0.97 3.57 ± 1.05 4.09 ± 1.98 4.17 ± 1.56
1.30 (0.92) 0.85 (0.60) 0.95 (0.56) 1.34 (0.81)
0.000** 0.015** 0.007** 0.000**
*Associated df = 10,10. †P values are for 2-tailed, paired sample t tests; associated df = 21. ‡Indicates significance at .05 level. §Associated df = 21,21.
Subject heart rate was calculated through use of a method similar to that outlined for the calculation of the CRI. Mean heart rates and SDs were calculated for each phase of the experiment through use of data from the last 20 seconds of the 5- to 7-minute resting period before any palpation and the full 2-minute data collection periods. Ninety-five percent confidence intervals (CIs) were constructed for the means and plotted for each subject. After inspection of the heart rate plots, we concluded that all subjects remained in a very similar state of arousal for the duration of the data collection. ICC (2,1) calculations, tests of significance, and determinations of Pearson product-moment coefficients (r) were performed.21-24 The ICC was used to indicate interexaminer and intraexaminer reliability; the ICC determines the extent of agreement, whereas the Pearson r is a measure of association rather than agreement.22,25 A custom-modeled factorial, 2-way analysis of variance (ANOVA) was used to assess for contributions of variance between subjects, examiners, positions, and trials and all 2-way interactions. Mean absolute differences in CRI rate were calculated for examiners at the head and the sacrum, and 95% CIs were also constructed. Two-tailed, paired sample t tests were used to assess for differences among the mean CRI rates for each examiner at the head and the sacrum. Pearson r values, 2tailed t test results, and ANOVA results were calculated through use of SPSS for Windows version 7.5 (SPSS Inc, Chicago, Ill). The criterion for statistical significance was set at the .05 level.
RESULTS The mean rate of the CRI palpated by examiner A across all subjects was 2.92 cycles.min–1 (SD = 0.49) at the head and 3.57 cycles.min–1 (SD = 0.54) at the sacrum. In contrast, the mean rate of the CRI palpated by examiner B across all subjects was 4.09 cycles.min–1 (SD = 1.01) at the head and 4.17 (SD = 0.80) cycles.min–1 at the sacrum (Table 1). The mean absolute differences in CRI rate palpated by the two examiners at the head and sacrum are displayed in Table 1.
Paired sample, 2-tailed t tests for differences in mean CRI rate recorded simultaneously by examiners at the head and the sacrum and for records taken consecutively in the same position were significantly different (Table 1). Plots of CRI rate as recorded by the 2 examiners palpating simultaneously at 2 different positions in the same subjects are presented in Figures 1 and 2. Similarly, plots of CRI rate as recorded by the 2 examiners at the same location (head or sacrum) in the same subject are presented in Figures 3 and 4. The results of the custom-modeled factorial ANOVA revealed significant differences between CRI rates recorded by the 2 examiners (F1,1 = 84.17, P < .001). There were also significant differences between CRI rates recorded in different positions (F1,1 = 14.58, P < .001) and significant differences between CRI rates recorded in different subjects (F10,10 = 7.65, P < .001). There were also significant 2-way interactions between examiner and position (F1,1 = 8.70, P = .005), examiner and subject (F1,10 = 5.98, P < .001), and examiner and trial (F1,1 = 4.45, P = .04). There were no significant interaction effects between position and subject (F1,10 = 1.80, P = .09), position and trial (F1,1 = 0.78; P = .38) or subject and trial (F10,1 = 1.55, P = .16). The ICC (2,1) reliability coefficients, Pearson r values, and results of associated tests of significance are presented in Table 1. Intraexaminer agreement was interpreted as being “fair to good,”23 the ICCs ranging from +0.47 (examiner A at the head) to +0.73 (examiner B at the head). Interexaminer agreement was poor to nonexistent, the ICCs ranging from –0.09 (examiner A at the head and examiner B at the sacrum) to +0.31 (examiner B at the head and examiner A at the sacrum).
DISCUSSION Statistics of Reliability Although the Pearson product-moment correlation coefficient (Pearson r) has previously been used to provide estimates of reliability, the use of this coefficient is considered inappropriate for the measurement of reliability.22,25,26 The Pearson r provides an index of association; however, in measuring reliability we wish to know the
187
188
Journal of Manipulative and Physiological Therapeutics Volume 24 • Number 3 • March/April 2001 Cranial Rhythmic Impulse • Moran and Gibbons
Table 2. Guidelines for ICC interpretation23
where n = 11. Although there are no clear, universally applicable, qualitative standards for interpreting ICCs, Fleiss23 provides a general guideline (Table 2).
between the subjects.27 Because the sample population displayed adequate heterogeneity, the negative ICC values for CRI rate palpation by the same examiner at different positions can be interpreted as single examiners recording CRI rates at the head that differed from those recorded at the sacrum in the same subject. The estimation of reliability in this and other studies is based on the assumption of CRI rate stability from trial to trial within the same subject. A real change in the rate of the CRI from one trial to another would lead to artificially low estimates of reliability. Unfortunately, there is no criterion standard for measurement of the CRI, and we were therefore unable to directly control for changes in CRI rate between trials. However, we did attempt to monitor the state of subject arousal over the course of data collection through measurement of subject heart rate. All subjects displayed uniform states of arousal over the course of data collection, as measured through use of heart rate. The variation in CRI rates as measured simultaneously in the same subject (Figures 1 and 2) tends to negate the need for demonstrating subject stability. If the rate of the CRI were unstable and varied across trials, then we would expect there to be low intraexaminer ICCs. This is evidently not the case (Table 1), and we suggest that a possible lack of stability of the CRI was not a contributing factor in the low interexaminer reliability coefficients recorded in this study.
Intraexaminer Reliability
Interexaminer Reliability
In our study, intraexaminer reliability for palpation of the CRI at the same position ranged from “fair to good,” significant ICCs ranging from +0.52 to +0.73 (Table 1). These estimates are considerably better than those reported by Rogers et al,14 who calculated ICC (2,1) coefficients between +0.18 and +0.30 for palpation of the CRI at the head and feet. We can offer little explanation for the differences in reliability between these 2 comparable studies. The 2 examiners in the study of Rogers et al had 5 and 17 years of experience using cranial techniques and had attended numerous courses and been the instructors in others. The examiners in our study had 4.5 and 6.5 years of experience and also had taken several postgraduate courses. Despite the opinion of Sibley et al,30 it is doubtful that any differences in the reported levels of experience can explain the improved intraexaminer reliability seen in our investigation in comparison with that of Rogers et al,14 especially given that the total level of examiner experience in the study of Rogers et al (combined examiners = 22 years) far exceeds the 11 years’ combined experience in the present study. ICCs for single examiners palpating CRI rate at the head and the sacrum were less than 0, reflecting no concordance between CRI rates as palpated by the same examiner at 2 positions in the same subject (Table 1). The theoretical lower limit of the ICC (2,1) might approach –∞.27 Such a case can occur when the difference between subjects approaches 0 and the measurement error is greater than 0. ICC values cannot be significant unless there are differences
Interexaminer reliability for palpation of the CRI ranged from poor to nonexistent , the ICCs ranging from –0.09 to +0.05 (Table 1). These coefficients reflect the very low levels of agreement that the 2 examiners had in the measurement of CRI rate. This was the case when examiners were simultaneously recording at opposite ends and when comparisons were made from examiners recording at the same position. These findings are displayed in Figures 1 through 4. Visual inspection of Figures 1 and 2 identifies numerous cases in which examiners simultaneously recorded different CRI rates at the head and the sacrum. At both the head and the sacrum, examiner A consistently palpated CRI rates that were lower than those palpated by examiner B. The low levels of interexaminer reliability reported in our study are comparable to those previously reported in the literature. Previous authors have reported interexaminer reliability coefficients (ICC [2,1]) ranging from –0.02 to +0.22.14,19,20 Clearly, there are differences in CRI rates palpated simultaneously from different positions. The “core-link” hypothesis of craniosacral motion predicts the synchronous movement of the occiput and the sacrum.10 It might therefore be expected that CRI rates recorded simultaneously at these positions would be in close agreement. Our results suggest that at least for the population of normal subjects whom we examined, this was not the case. The results of this study and previous published data14 raise doubts as to the robustness of the construct validity for the “core-link” concept as proposed by Sutherland1 and Magoun.3
ICC value
Strength of concordance
<0.4 0.4-0.75 >0.75
Poor Fair to good Excellent
extent to which 2 measurements yield the same result. This is known as concordance. 22 The ICC provides such an index and is the extent of agreement of one measurement on a subject and another measurement on the same subject.21 One assumption that must be met for valid use of ICCs is that of subject heterogeneity. A reduced level of variation among samples can contribute to artificially low reliability coefficients calculated through use of ICCs.27,28 In our study, we assumed that the sample population would present with a range of different CRI rates. The presence of differences between subjects (F = 7.65, df = 10, P = .000) suggests that the assumption of subject heterogeneity was met. A second assumption for valid use of ICCs is that of sample population size.29 The number of subjects (n) should be large enough that n/(n – 1) is close to 1. Our study meets this assumption because n/(n – 1) = 1.1,
Journal of Manipulative and Physiological Therapeutics Volume 24 • Number 3 • March/April 2001 Cranial Rhythmic Impulse • Moran and Gibbons
Although the low levels of interexaminer agreement raise doubts as to the construct validity of the theoretical “corelink” model, the reliability data should also be interpreted in a more clinical context. Di Fabio31 writes that the study of reliability should be accompanied by an explanation of why the measure is important and what influence the measurement has on determining the nature and scope of treatment. One question arising from the interexaminer data is this: Would the palpation of different CRI rates lead the examiners to different diagnostic conclusions and therefore raise the possibility of different and possibly even opposite interventions? Although we did not design our experiment to specifically answer such a question, an indication might be possible on the basis of inspection of the rate plots (Figures 1-4) and consideration of the calculated interexaminer correlation data (Table 1). There were low to nonexistent levels of correlation between examiners palpating simultaneously at the head and the feet (and vice versa) and between examiners consecutively palpating at the same position (nonsignificant r coefficients ranging from –0.314 to +0.45). Inspection of the data for subject 7 in Figure 3 reveals an occasion when one of the examiners palpated a CRI rate (2.8 cycles.min–1) that some proponents would consider to be low or even to be associated with significant pathosis.5,32,33 The second examiner palpating the same subject within minutes of the first recorded a CRI rate (6.07 cycles.min–1) that some proponents would consider to be within the “normal” range.2 In a clinical environment, such discrepancy might have led examiner A to a clinical impression different from that reached by examiner B. In contrast, inspection of Figures 3 and 4 reveals occasions when the examiners were in close agreement with respect to CRI rate. At the head the examiners were in close agreement as to CRI rates for subjects 4, 8, and 9 (Figure 3), whereas at the sacrum the examiners agreed within 1 cycles.min–1 for subjects 4, 5, 6, 8, 9, and 10 (Figure 4). The mean absolute differences between CRI rates recorded by the 2 examiners at various positions are all less than 1.4 cycles.min–1 (Table 2), and though these differences are statistically significant, the clinical importance of the results are less clear. Because a diagnosis of dysfunction is unlikely to be based on assessment of CRI rate alone, reliability studies in the cranial arena should be extended to include other clinically important diagnostic parameters.
CRI Rate Comparisons The mean CRI rates as palpated by the 2 examiners at the head and the sacrum ranged from 2.92 (SD = 0.49) to 4.17 (SD = 0.80) cycles.min–1 (Table 1). The highest single rate recorded was 6.07 cycles.min–1 and the lowest was 2.17 cycles.min–1. These rates are comparable to those recorded in reliability studies.14,20 Wirth-Pattullo and Hayes19 reported slightly faster CRI rates in their reliability study, the mean rates ranging from 4.5 to 7 cycles.min–1. There is little consensus in the literature as to what CRI rate should be considered “normal.”34 Various authors report a range of CRI rates—from 0.6 cycles.min–1 (the slowest) to 10 to 14 cycles.min–1.11,35
That the CRI is generated as a harmonic frequency derived from “multiple biological oscillators,”36 or as a complex interaction of tissue fluid pressure dynamics involving examiner and subject37 is an interesting hypothesis for consideration. Recently, a new hypothesis for the origin of the CRI has been proposed in which the supposed CRI might be felt as an expression of venous vessel wall pulsation, or “venomotion.”38 If future experimental research were to provide evidence for models of CRI generation based on the complex interaction of multiple frequencies (cardiac, respiratory, venous, and so forth) originating from examiner and/or subject, then comparisons of CRI rate between different examiners might become meaningless. In any case, such a finding does not preclude the pressing need for investigation into clinically measurable patient outcomes after treatment through use of cranial techniques. It should be noted that the sample size in this study was small, and some caution should be used in extrapolating the results to the population of cranial practitioners in general. Future studies would benefit from the use of a larger stratified sample of cranial practitioners that would more accurately reflect the age and experience of the cranial practitioner population.
CONCLUSION This study aimed to (1) estimate coefficients of intraexaminer and interexaminer reliability for examiners palpating what they recognized as the CRI, (2) investigate the construct validity of the “core-link” hypothesis, and (3) compare the mean CRI rates recorded by 2 examiners with those reported in the previous literature. Intrarater reliability for the examiners at either the head or the sacrum was fair to good, significant ICCs ranging from +0.52 to +0.73. Interexaminer reliability for simultaneous palpation at the head and the sacrum was poor to nonexistent, the ICCs ranging from –0.09 to +0.31. There were significant differences between rates of the CRI palpated simultaneously at the head and the sacrum. These data fail to support the construct validity of the “core-link” hypothesis as traditionally held by proponents of craniosacral therapy and OCF. The mean CRI rates reported in this study range between 2.92 and 4.17 cycles.min–1 and are comparable to rates reported in similar experimental studies. These mean rates might be viewed by some practitioners as lower than “normal.” Future research should focus on establishing the origin of the CRI phenomenon, further evaluate the reliability of cranial examination procedures (including other diagnostic parameters), and investigate measurable clinical outcomes from such interventions.
ACKNOWLEDGMENTS We thank Ian Fairweather of the Faculty of Human Movement, Victoria University, for providing technical support and the custom designed software application used in this study. We also thank the osteopaths who participated as examiners in this study.
189
190
Journal of Manipulative and Physiological Therapeutics Volume 24 • Number 3 • March/April 2001 Cranial Rhythmic Impulse • Moran and Gibbons
REFERENCES 1. Sutherland WG. The cranial bowl. Minnesota: Free Press Company; 1939. 2. Upledger JE, Vredevoogd JD. Craniosacral therapy. Seattle: Eastland Press; 1983. p. 16. 3. Magoun HI. Osteopathy in the cranial field. Kirksville (MO): Journal Printing Company; 1976. p. 18-9. 4. Wales AL, editor. Teachings in the science of osteopathy. Rudra Press; 1990:22,66,165. 5. Woods JM, Woods RH. A physical finding related to psychiatric disorders. J Am Osteopath Assoc 1961:60:988-93. 6. Sackett DL, Haynes RB, Tugwell P. Clinical epidemiology: a basic science for clinical medicine. 2nd ed. Boston: Little, Brown; 1991. p. 19-49. 7. Vincent-Smith B, Gibbons P. Inter-examiner and intra-examiner reliability of the standing flexion test. Manual Ther 1999; 4:87-93. 8. Bogduk N. An interview with Nicolai Bogduk: musculoskeletal medicine, evidence based medicine and the meaning of life. Newsletter of the Australian Association of Musculoskeletal Medicine 1998. p. 1-4. 9. Leboeuf-Yde C. How real is the subluxation? a research perspective. J Manipulative Physiol Ther 1998;21:492-4. 10. Lay EM. Cranial field. In: Ward RC, editor. Foundations of osteopathic medicine. Baltimore: Williams and Wilkins; 1997. p. 905. 11. Chaitow L. Cranial manipulation theory and practice. Edinburgh: Churchill Livingstone; 1999. p. 22,27. 12. Ettlinger H, Gintis B. Cranial concepts. In: Di Giovanna EL, Schiowitz S, editors. An osteopathic approach to diagnosis and treatment. 2nd ed. Philadelphia: Lippincott; 1991. p. 405. 13. Zanakis MF, DiMeo J, Madonna BFA, Morgan BA, Drasby BA. Objective measurement of the CRI with manipulation and palpation of the sacrum [abstract]. J Am Osteopath Assoc 1996;96:551. 14. Rogers JS, Witt PL, Gross MT, Hacke JD, Genova PA. Simultaneous palpation of the craniosacral rate at the head and feet: intrarater and interrater reliability and rate comparisons. Phys Ther 1998;78:1175-85. 15. Norton JA. Challenge to the concept of craniosacral interaction. Acad Applied Osteopathy J 1996;6:15-21. 16. Chaitow L. Cranial manipulation re-examined. Intern J Alternative Complementary Med 1997. p. 29-32. 17. Butler D. Mobilization of the nervous system. Edinburgh: Churchill Livingstone; 1991. 18. Upledger JE. The reproducibility of craniosacral examination findings: a statistical analysis. J Am Osteopath Assoc 1977; 76:890-9.
19. Wirth-Pattullo V, Hayes KW. Interrater reliability of craniosacral rate measurements and their relationship with subjects’ and examiners’ heart and respiratory rate measurements. Phys Ther 1994. p. 908-20. 20. Hanten WP, Dawson DD, Iwata M, Seiden M, Whitten FG, Zink T. Craniosacral rhythm: reliability and relationships with cardiac and respiratory rates. J Orthop Sports Phys Ther 1998;27:213-8. 21. Shrout PE, Fleiss JL. Intraclass correlations: uses in assesing rater reliability. Psychol Bull 1979;86:420-8. 22. Kramer MS, Feinstein AR. Clinical biostatistics: the biostatistics of concordance. Clin Pharmacol Ther 1981;29:111-23. 23. Fleiss JL. The design and analysis of clinical experiments. New York: John Wiley and Sons; 1986. p. 7. 24. Bartko JJ. On various intraclass correlation reliability coefficients. Psychol Bull 1976;83:762-5. 25. Haas M. Statistical methodology for reliability studies. J Manipulative Physiol Ther 1991;14:119-32. 26. Haas M. The reliability of reliability. J Manipulative Physiol Ther 1991;14:199-208. 27. Lahey MA, Downey RG, Saal FE. Intraclass correlations: there’s more than meets the eye. Psychol Bull 1983;93:586-95. 28. Mitchell SK. Interobserver agreement, reliability, and generalizability of data collected in observational studies. Psychol Bull 1979;86:376-90. 29. Fleiss JL. Measuring agreement between two judges on the presence or absence of a trait. Biometrics 1975;31:651-9. 30. Sibley G, Broder-Oldach RE, Norton JM. Interexaminer agreement in the characterization of the cranial rhythmic impulse [abstract]. J Am Osteopath Assoc 1992;92:1285. 31. Di Fabio RP. Studies of reliability: show me the question [editorial]. J Orthop Sports Phys Ther 1999;29:370-1. 32. Karni Z, Upledger JE, Mirzahi J, Heller L, Becker E, Najenson T. Examination of the cranial rhythm in long standing coma and chronic neurological cases. In: Upledger JE, Vredevoogd JD. Craniosacral therapy. Seattle: Eastland Press; 1983:275-81. 33. Brookes D. Lectures on cranial osteopathy. Wellingborough, United Kingdom: Thorsons Publishers; 1981. p. 56-9. 34. Chaitow L. Cranial dogma re-assessed. Intern J Alternative Complementary Med 1997. p. 28-32. 35. Greenman PE. Principles of manual medicine. 2nd ed. Baltimore: Williams and Wilkins; 1996. p. 165. 36. McPartland JM, Mein EA. Entrainment and the cranial rhythmic impulse. Alternative Therapies 1997;3:40-5. 37. Norton JM. A tissue pressure model for the palpatory perception of the cranial rhythmic impulse. J Am Osteopath Assoc 1991;91:975-94. 38. Farasyn A. New hypothesis for the origin of cranio-sacral motion. J Bodywork Movement Ther 1999;3:229-37.
Intraexaminer and Interexaminer Reliability for Palpation of the Cranial Rhythmic Impulse at the Head and Sacrum Robert W. Moran,a and Peter Gibbons, MB, BS, DO, DM-SMedb
ABSTRACT Background: A range of health care practitioners use cranial techniques. Palpation of a cranial rhythmic impulse (CRI) is a fundamental clinical skill used in diagnosis and treatment with these techniques. There has been little research establishing the reliability of CRI rate palpation. Objective: This study aimed to establish the intraexaminer and interexaminer reliability of CRI rate palpation and to investigate the “core-link” hypothesis of craniosacral interaction that is used to explain simultaneous motion at the cranium and sacrum. Design: Within-subjects, repeated-measures design. Subjects: Two registered osteopaths, both with postgraduate training in diagnosis and treatment, using cranial techniques, palpated 11 normal healthy subjects. Methods: Examiners simultaneously palpated for the CRI at the head and the sacrum of each subject. Examiners indicated the “full flexion” phase of the CRI by activating silent foot
INTRODUCTION Various forms of cranial manipulation are practiced by a variety of practitioners, including chiropractors, osteopaths, physical therapists, and massage therapists. These techniques have been variously referred to as osteopathy in the cranial field (OCF), cranial osteopathy, craniosacral therapy, sacrooccipital technique, and somatic cranial work. Of these, osteopathy in the cranial field, as first described by W.G. Sutherland1 in 1939, and craniosacral therapy, as popularized by Upledger2 since the late 1970s, are probably the most commonly practiced. For the purposes of this paper, OCF and craniosacral therapy will be considered together because both rely on the concept of involuntary cranial and sacral bone motion and on palpation of a rhythm or impulse in diagnosis a Lecturer, School of Health Sciences, Victoria University, Melbourne, Australia. b Associate Professor, School of Health Sciences, Faculty of Human Development, City Campus, Victoria University, Melbourne, Australia. Submit reprint requests to: Peter Gibbons, MB, BS, DO, DMSMed, School of Health Sciences, Faculty of Human Development, City Campus, Victoria University, PO Box 14428 MC, Melbourne 8001, Australia. E-mail: [email protected]. Paper submitted January 19, 2000; in revised form March 24, 2000.
doi:10.1067/mmt.2001.113773
switches that were interfaced with a computer. Subject arousal was monitored using heart rate. Examiners were blind to each other’s results and could not communicate during data collection. Results: Reliability was estimated from calculation of intraclass correlation coefficients (2,1). Intrarater reliability for examiners at either the head or the sacrum was fair to good, significant intraclass correlation coefficients ranging from +0.52 to +0.73. Interexaminer reliability for simultaneous palpation at the head and the sacrum was poor to nonexistent, ICCs ranging from –0.09 to +0.31. There were significant differences between rates of CRI palpated simultaneously at the head and the sacrum. Conclusions: The results fail to support the construct validity of the “core-link” hypothesis as it is traditionally held by proponents of craniosacral therapy and osteopathy in the cranial field. (J Manipulative Physiol Ther 2001;24:183-90) Key Indexing Terms: Palpation; Reliability; Osteopathic Manipulation; Osteopathy
and treatment of dysfunction. Likewise, the terms primary respiratory mechanism,1 craniosacral motion/rhythm,1 primary respiratory impulse,2 Sutherland wave,3 the Tide,4 and cranial rhythmic impulse (CRI),5 will be used synonymously. The existence of the CRI, as described by Sutherland1 and others, is yet to be definitively demonstrated through use of modern instrumentation. Because there is currently no objective criterion standard or instrumentation of known validity that can be used to measure the CRI, subjective palpation for CRI characteristics persists as the only diagnostic method used in clinically evaluating this rhythm. This lack of objective measurement makes the issue of reliability of palpation as a diagnostic tool important both to clinicians in private practice and to those who teach these techniques. Because of this, great emphasis is placed in teaching skills for CRI palpation to beginners in both postgraduate and undergraduate cranial technique courses. Despite the apparent importance of the role of palpation in diagnosis and treatment, there has been little published research that examines reliability or validity for palpation of CRI rate or indeed of any other clinical characteristic, such as the site or nature of the perceived dysfunction. Likewise, the lack of a criterion standard for CRI measurement means that the validity of palpation as a measure of the CRI has not been established.
183
184
Journal of Manipulative and Physiological Therapeutics Volume 24 • Number 3 • March/April 2001 Cranial Rhythmic Impulse • Moran and Gibbons
Importance of Reliability Using diagnostic methods of known reliability is important for accurate clinical decision-making.6-8 Both Bogduk8 and Leboeuf-Yde9 make persuasive arguments for establishing the reliability of diagnostic measures and implementing these tests in clinical practice. Their arguments are summarized in the following points: 1. In an educational environment, the time and money spent on teaching and learning techniques that remain unvalidated might not be justifiable. 2. In clinical practice, the application of many therapeutic techniques is based on the results of diagnostic procedures; these procedures should be validated and reliable. 3. Subsequent therapeutic decisions can be confounded by inappropriate patient allocation for different interventions, thereby compromising patient outcomes. 4. The notion of preventive care in asymptomatic patients might be unjustifiable if such care is based on unvalidated and unreliable diagnostic procedures.
Concepts of Craniosacral Interaction One of the central concepts in both OCF and craniosacral therapy is that of rhythmic involuntary motion of the sacrum related to the involuntary motion of the occiput. Sutherland was the first to propose such a link between the occiput and sacrum.4 Magoun3 named this the “core-link” and described an involuntary movement of the sacrum caused by a lifting force exerted on the sacrum by the attachments of the spinal dura to the sacrum. Magoun proposed that the effect of this was to rotate the sacrum around a transverse axis at the level of S2. The lifting force is apparently provided by the anterosuperior movement of the foramen magnum during the flexion phase of the CRI. Lay10 reports that the involuntary motion of the sacrum is synchronous with the involuntary motion of the occiput. The concept of the link between occiput and sacrum is widely recognized by proponents of both craniosacral therapy and OCF.2,10-12 Despite the apparent importance of this concept, there has been only one published experimental study that provides any supportive evidence for its existence. Zanakis et al13 recorded movement of surface markers on skin overlying cranial bones while an experienced practitioner simultaneously palpated apparent sacral motion and indicated “flexion” using a foot switch. Zanakis et al reported a 92% level of percentage agreement and concluded that their findings provided support for a hypothesized synchronous craniosacral relationship. Two recent studies have presented data that call the “corelink” concept into question. Rogers et al14 and Norton15 conducted palpation studies in which two examiners were able to palpate for the CRI in the same subject simultaneously from different positions. Norton’s examiners were positioned at the head and sacrum, whereas in the study of Rogers et al the examiners were at the head and feet. Both studies report low levels of agreement between simultaneous findings at the two positions. Chaitow11,16 presents an argument against the linking hypothesis. Chaitow cites the
clinical observations of Butler,17 who reports that to clinically identify adverse neural tension within the neuroaxis, slack must be taken up by placing the patient in a “slump position” of full neck and trunk flexion. Chaitow argues that in the light of such observation, “the core-link hypothesis is probably seriously flawed.”11
Previous Reliability Studies for CRI Rate Upledger18 investigated the reliability of pairs of examiners performing craniosacral examinations on 25 preschool children. Although CRI rate was one of the primary measures of the examination, the results were not published. Subsequent statistical analysis of Upledger’s raw CRI rate data by Wirth-Pattullo and Hayes19 resulted in the calculation of an intraclass correlation coefficient (ICC) of +0.57. However, Upledger’s report fails to identify the method by which CRI rate was measured. This omission makes it impossible to adequately critique Upledger’s conclusions that acceptable interexaminer reliability was attained and that this lends “considerable evidence to the existence of a real and perceptible craniosacral motion system.”18 On the basis of the method as presented, we have difficulty reconciling Upledger’s results with his conclusions. Wirth-Pattullo and Hayes19 conducted an interrater reliability study with 3 physical therapists trained in craniosacral therapy. Each examiner independently palpated the heads of 12 children and adults. Heart and respiratory rates of both examiners and subjects were manually recorded. Correlations between craniosacral rate and subject and examiner heart and respiratory rates were low and not statistically significant. Interrater reliability of craniosacral rate was estimated from calculation of the ICC.2,1 The authors conclude that the resulting reliability coefficient of –0.02 was unacceptable for clinical decision-making. Hanten et al20 sought to determine interexaminer and intraexaminer reliability for palpation of the CRI at the heads of 40 healthy adults. Hanten et al also recorded heart and respiratory rates of examiners and subjects. Two examiners, each with 9 months of experience in palpating craniosacral rhythm, achieved excellent intraexaminer reliability coefficients (ICC [1,1]) of +0.78 and +0.83. However, the interexaminer reliability coefficient of +0.22 was poor. Interrater and intrarater reliability for the measurement of CRI rate was also investigated by Rogers et al.14 They used a repeated-measures design whereby each of 2 trained examiners palpated for the CRI simultaneously in the same subject (n = 28). Examiners activated foot switches in time with palpated events to enable calculation of a CRI rate. One examiner would palpate at the head while the other examiner palpated at the feet. The authors report interrater reliability coefficients (ICC [2,1]) of +0.08 at the head and +0.19 at the feet. Intrarater ICCs range from +0.18 to +0.30, and craniosacral rates recorded simultaneously at the head and feet are reported as being different. Rogers et al conclude that their results suggest that examiners were measuring different phenomena and that they may have been attempting to measure something that did not exist.
Journal of Manipulative and Physiological Therapeutics Volume 24 • Number 3 • March/April 2001 Cranial Rhythmic Impulse • Moran and Gibbons
Objectives and Purposes The purposes of this study were as follows: 1. To examine the interexaminer and intraexaminer reliability of measurement of CRI rate by palpation 2. To examine the relationship between CRI measurements recorded simultaneously at the head and sacrum by 2 examiners (we examined construct validity of the “corelink” theory that the CRI at the sacrum should be the same as that at the head, and we sought to test the null hypothesis that there should be no difference in CRI rates recorded by 2 examiners palpating independently but simultaneously at the head and sacrum) 3. To document the rate of the CRI as palpated by 2 examiners in normal healthy subjects (n = 11) and compare these findings with those reported in the literature.
METHODS Participants A convenience sample of 11 healthy volunteers was recruited by the investigators through word of mouth. Five subjects were male and 6 female. The mean age was 25.8 (SD = 8.0) years and the range 18-44 years. To be eligible to participate, each subject had to be between 18 and 80 years of age and able to lie quietly supine on a treatment table for 45 minutes and could not currently be undergoing any manual therapy or medical treatment. At the time of the study, none of the subjects were experiencing any pain or discomfort that would lead them to seek health care advice or treatment. The study was approved by the Victoria University Human Research Ethics Committee. The procedure was explained to the subjects before it was begun, and they were free to withdraw from the study at any stage. All subjects read and signed informed consent forms. All subjects completed the study.
procedures at the alternate positions. Examiners at the head and sacrum were separated by a large, triple-thickness curtain that hung across the subject at the level of the chest. This design allowed for the collection of 2 independent recordings, taken simultaneously at 2 different positions on the same subject (eg, with examiner A palpating at the head, examiner B simultaneously palpated at the sacrum). Heart rate was recorded in an attempt to monitor the state of arousal of the subject from trial to trial. The procedure was designed to allow for analysis of interexaminer and intraexaminer reliability.
Variables The dependent variable in the study was rate of the CRI, the units being cycles.min–1; the independent variables were examiner (A or B), position (head or sacrum), and measurement trial (1 or 2). For each of the 11 subjects, 8 CRI rate measurements were calculated, as follows: examiner A at the head, trial 1; examiner B at the sacrum, trial 1; examiner A at the head, trial 2; examiner B at the sacrum, trial 2; examiner A at the sacrum, trial 1, examiner B at the head, trial 1; examiner A at the sacrum, trial 2, and examiner B at the head, trial 2.
Instrumentation
Two Victorian State–registered osteopaths participated in the study as examiners. At the time of the study, examiner A had been in private practice for a period of 4.5 years and had completed 2 Sutherland Cranial Teaching Foundation (SCTF) basic-level courses. Examiner A estimated that 100% of patients under their care would undergo at least some cranial treatment and that approximately 30% to 40% of patients would undergo cranial treatment as the major treatment modality. Examiner B had been in private practice for a period of 6.5 years and had completed 2 SCTF basiclevel courses and 1 SCTF intermediate-level course. Examiner B estimated that nearly 100% of patients under their care would undergo at least some cranial treatment and that nearly 100% of patients would receive cranial treatment as the major treatment modality.
Two momentary-action foot switches (model 316-901, RS Components Pty Ltd, VIC, Australia) were activated in time with the designated “full flexion” phase of the CRI as palpated by the examiners. To blind examiners to any auditory cues, the foot switches were positioned inside soundproof housings and a computer fan provided a source of white noise in the background. The operation of the foot switches in the housings was inaudible. Sufficient time was allocated for the examiners to practice activation of the foot switches and become comfortable with their use. Evaluation of correct operation of the foot switches was undertaken before the commencement of formal data collection. Examiners activated their foot switches in response to a visual stimulus presented simultaneously to both examiners. In 3 trials, the 2 examiners were 100% reliable in responding to the visual stimuli and demonstrated perfect interexaminer agreement (for trial 1, ICC [2,1] = 1.0, F5,5 = 5835.84 , and P < .001; for trial 2, ICC [2,1] = 1.0, F8,8 = 9711.36,and P < .001; for trial 3: ICC [2,1] = 1.0, F6,6 = 13574.57, and P < .001). Each subject wore a standard heart rate monitor chest strap (Polar X-trainer Plus, Polar Electro Oy, Finland) throughout the study. Heart rate data were relayed via telemetry to a transceiver positioned near the subject. A custom-designed computer software program (LabVIEW version 5.01, National Instruments, Austin, Tex) was used to record subject heart rate and examiner foot switch data. Data were captured by means of an analog-to-digital board connected to an IBM-compatible personal computer.
Outline of Study Design
Procedures
The study used a within-subjects, repeated-measures design, as previously outlined by Rogers et al.14 Each examiner palpated each subject at the head or the sacrum for 2 trials before the examiners exchanged positions and repeated the
Before the study, the 2 examiners met with the investigator to define the event that was to be recorded by the activation of the foot switch. After discussion, the examiners agreed that they would activate the switch at the
Examiners
185
186
Journal of Manipulative and Physiological Therapeutics Volume 24 • Number 3 • March/April 2001 Cranial Rhythmic Impulse • Moran and Gibbons
Fig 1. Simultaneous CRI rates for examiner A at head and examiner
Fig 3. CRI rates for both examiners at head. Each value is mean ± SE.
B at sacrum. Each value is mean ± SE.
Fig 2. Simultaneous CRI rates for examiner B at head and examiner A at sacrum. Each value is mean ± SE.
instant they recognized the palpatory sensation of the CRI phase described as “full flexion.” Examiners were free to nominate the palpation contact of their own choosing and to vary this contact between subjects according to their own judgment. The subject was positioned supine on a standard treatment table with the heart rate monitor chest strap attached. The curtain was positioned across the subject and secured. The subject rested quietly for 5 to 7 minutes to allow the heart rate to become stable before commencement of data collection. After this period, the examiners were instructed to take up their contacts and were allowed sufficient time to establish a strong sense of the CRI before any data were collected. When the examiners expressed confidence that they had a sense of the CRI, they independently indicated to the investigator that they were ready to commence formal data collection. When both examiners were ready, a 2minute trial began with an instruction to the examiners to “Start data collection.” During this period, the examiners activated their respective foot switches at the moment they sensed the “full flexion” phase of the CRI. After 2 minutes, the investigator announced “Stop palpation” and the examiners released their contacts and relaxed for a minimum of 2 minutes. After the initial trial, there was a second 2-minute
Fig 4. CRI rates for both examiners at sacrum. Each value is mean (± SE). trial with the examiners positioned in the same locations. After 2 trials, the examiners exchanged positions before commencing 2 more trials in the alternate positions. Each examiner thus undertook four 2-minute trials (2 at the head and 2 at the sacrum) per subject. Each trial was of 2 minutes’ duration, and there was a minimum rest period of 2 minutes between trials. Examiner starting positions were alternated between consecutive subjects to avoid any systematic error. Data were collected over a 6-hour period. The examiners had 5-minute rest periods between subjects, and neither examiner felt unduly fatigued by the procedure.
Data Reduction and Statistical Analysis Raw data from the CRI data acquisition software were exported to Microsoft Excel (Microsoft, Redmond, Wash) for descriptive analysis. CRI rates were calculated from the interval times between events recorded from the activation of each examiner’s foot switch. This method resulted in the calculation of instantaneous CRI rates for each trial. The mean CRI rate for each trial was calculated from the instantaneous rates. Mean CRI rates for each subject were calculated from the pooled trial data in each position.
Journal of Manipulative and Physiological Therapeutics Volume 24 • Number 3 • March/April 2001 Cranial Rhythmic Impulse • Moran and Gibbons
Table 1. Results for palpation of CRI rate at head and sacrum for both examiners ICC (2,1) coefficient Intraexaminer Examiner A at head Examiner A at sacrum Examiner B at head Examiner B at sacrum Examiner A at head and sacrum Examiner B at head and sacrum Interexaminer Examiner A at head, examiner B at sacrum Examiner B at head, examiner A at sacrum Both examiners at sacrum Both examiners at head
F value*
Pearson r coefficient
P value
+0.47 +0.65 +0.73 +0.52 –1.46 –1.47
2.61 5.17‡ 8.35‡ 4.71‡ –0.03§ –0.07‡
+0.47 +0.68 +0.80 +0.74 +0.31 +0.62
0.144 0.022‡ 0.003‡ 0.009‡ 0.164 0.002‡
–0.09 +0.31 –0.02 +0.05
0.54 2.18 0.93 1.27
–0.314 +0.45 –0.035 +0.17
0.320 0.158 0.920 0.640
CRI mean rate ± 95% CI (cycles.min–1)
CRI mean rate (SD) absolute difference (cycles.min–1)
Difference in mean CRI rate P value†
2.92 ± 0.97 3.57 ± 1.05 4.09 ± 1.98 4.17 ± 1.56
1.30 (0.92) 0.85 (0.60) 0.95 (0.56) 1.34 (0.81)
0.000** 0.015** 0.007** 0.000**
*Associated df = 10,10. †P values are for 2-tailed, paired sample t tests; associated df = 21. ‡Indicates significance at .05 level. §Associated df = 21,21.
Subject heart rate was calculated through use of a method similar to that outlined for the calculation of the CRI. Mean heart rates and SDs were calculated for each phase of the experiment through use of data from the last 20 seconds of the 5- to 7-minute resting period before any palpation and the full 2-minute data collection periods. Ninety-five percent confidence intervals (CIs) were constructed for the means and plotted for each subject. After inspection of the heart rate plots, we concluded that all subjects remained in a very similar state of arousal for the duration of the data collection. ICC (2,1) calculations, tests of significance, and determinations of Pearson product-moment coefficients (r) were performed.21-24 The ICC was used to indicate interexaminer and intraexaminer reliability; the ICC determines the extent of agreement, whereas the Pearson r is a measure of association rather than agreement.22,25 A custom-modeled factorial, 2-way analysis of variance (ANOVA) was used to assess for contributions of variance between subjects, examiners, positions, and trials and all 2-way interactions. Mean absolute differences in CRI rate were calculated for examiners at the head and the sacrum, and 95% CIs were also constructed. Two-tailed, paired sample t tests were used to assess for differences among the mean CRI rates for each examiner at the head and the sacrum. Pearson r values, 2tailed t test results, and ANOVA results were calculated through use of SPSS for Windows version 7.5 (SPSS Inc, Chicago, Ill). The criterion for statistical significance was set at the .05 level.
RESULTS The mean rate of the CRI palpated by examiner A across all subjects was 2.92 cycles.min–1 (SD = 0.49) at the head and 3.57 cycles.min–1 (SD = 0.54) at the sacrum. In contrast, the mean rate of the CRI palpated by examiner B across all subjects was 4.09 cycles.min–1 (SD = 1.01) at the head and 4.17 (SD = 0.80) cycles.min–1 at the sacrum (Table 1). The mean absolute differences in CRI rate palpated by the two examiners at the head and sacrum are displayed in Table 1.
Paired sample, 2-tailed t tests for differences in mean CRI rate recorded simultaneously by examiners at the head and the sacrum and for records taken consecutively in the same position were significantly different (Table 1). Plots of CRI rate as recorded by the 2 examiners palpating simultaneously at 2 different positions in the same subjects are presented in Figures 1 and 2. Similarly, plots of CRI rate as recorded by the 2 examiners at the same location (head or sacrum) in the same subject are presented in Figures 3 and 4. The results of the custom-modeled factorial ANOVA revealed significant differences between CRI rates recorded by the 2 examiners (F1,1 = 84.17, P < .001). There were also significant differences between CRI rates recorded in different positions (F1,1 = 14.58, P < .001) and significant differences between CRI rates recorded in different subjects (F10,10 = 7.65, P < .001). There were also significant 2-way interactions between examiner and position (F1,1 = 8.70, P = .005), examiner and subject (F1,10 = 5.98, P < .001), and examiner and trial (F1,1 = 4.45, P = .04). There were no significant interaction effects between position and subject (F1,10 = 1.80, P = .09), position and trial (F1,1 = 0.78; P = .38) or subject and trial (F10,1 = 1.55, P = .16). The ICC (2,1) reliability coefficients, Pearson r values, and results of associated tests of significance are presented in Table 1. Intraexaminer agreement was interpreted as being “fair to good,”23 the ICCs ranging from +0.47 (examiner A at the head) to +0.73 (examiner B at the head). Interexaminer agreement was poor to nonexistent, the ICCs ranging from –0.09 (examiner A at the head and examiner B at the sacrum) to +0.31 (examiner B at the head and examiner A at the sacrum).
DISCUSSION Statistics of Reliability Although the Pearson product-moment correlation coefficient (Pearson r) has previously been used to provide estimates of reliability, the use of this coefficient is considered inappropriate for the measurement of reliability.22,25,26 The Pearson r provides an index of association; however, in measuring reliability we wish to know the
187
188
Journal of Manipulative and Physiological Therapeutics Volume 24 • Number 3 • March/April 2001 Cranial Rhythmic Impulse • Moran and Gibbons
Table 2. Guidelines for ICC interpretation23
where n = 11. Although there are no clear, universally applicable, qualitative standards for interpreting ICCs, Fleiss23 provides a general guideline (Table 2).
between the subjects.27 Because the sample population displayed adequate heterogeneity, the negative ICC values for CRI rate palpation by the same examiner at different positions can be interpreted as single examiners recording CRI rates at the head that differed from those recorded at the sacrum in the same subject. The estimation of reliability in this and other studies is based on the assumption of CRI rate stability from trial to trial within the same subject. A real change in the rate of the CRI from one trial to another would lead to artificially low estimates of reliability. Unfortunately, there is no criterion standard for measurement of the CRI, and we were therefore unable to directly control for changes in CRI rate between trials. However, we did attempt to monitor the state of subject arousal over the course of data collection through measurement of subject heart rate. All subjects displayed uniform states of arousal over the course of data collection, as measured through use of heart rate. The variation in CRI rates as measured simultaneously in the same subject (Figures 1 and 2) tends to negate the need for demonstrating subject stability. If the rate of the CRI were unstable and varied across trials, then we would expect there to be low intraexaminer ICCs. This is evidently not the case (Table 1), and we suggest that a possible lack of stability of the CRI was not a contributing factor in the low interexaminer reliability coefficients recorded in this study.
Intraexaminer Reliability
Interexaminer Reliability
In our study, intraexaminer reliability for palpation of the CRI at the same position ranged from “fair to good,” significant ICCs ranging from +0.52 to +0.73 (Table 1). These estimates are considerably better than those reported by Rogers et al,14 who calculated ICC (2,1) coefficients between +0.18 and +0.30 for palpation of the CRI at the head and feet. We can offer little explanation for the differences in reliability between these 2 comparable studies. The 2 examiners in the study of Rogers et al had 5 and 17 years of experience using cranial techniques and had attended numerous courses and been the instructors in others. The examiners in our study had 4.5 and 6.5 years of experience and also had taken several postgraduate courses. Despite the opinion of Sibley et al,30 it is doubtful that any differences in the reported levels of experience can explain the improved intraexaminer reliability seen in our investigation in comparison with that of Rogers et al,14 especially given that the total level of examiner experience in the study of Rogers et al (combined examiners = 22 years) far exceeds the 11 years’ combined experience in the present study. ICCs for single examiners palpating CRI rate at the head and the sacrum were less than 0, reflecting no concordance between CRI rates as palpated by the same examiner at 2 positions in the same subject (Table 1). The theoretical lower limit of the ICC (2,1) might approach –∞.27 Such a case can occur when the difference between subjects approaches 0 and the measurement error is greater than 0. ICC values cannot be significant unless there are differences
Interexaminer reliability for palpation of the CRI ranged from poor to nonexistent , the ICCs ranging from –0.09 to +0.05 (Table 1). These coefficients reflect the very low levels of agreement that the 2 examiners had in the measurement of CRI rate. This was the case when examiners were simultaneously recording at opposite ends and when comparisons were made from examiners recording at the same position. These findings are displayed in Figures 1 through 4. Visual inspection of Figures 1 and 2 identifies numerous cases in which examiners simultaneously recorded different CRI rates at the head and the sacrum. At both the head and the sacrum, examiner A consistently palpated CRI rates that were lower than those palpated by examiner B. The low levels of interexaminer reliability reported in our study are comparable to those previously reported in the literature. Previous authors have reported interexaminer reliability coefficients (ICC [2,1]) ranging from –0.02 to +0.22.14,19,20 Clearly, there are differences in CRI rates palpated simultaneously from different positions. The “core-link” hypothesis of craniosacral motion predicts the synchronous movement of the occiput and the sacrum.10 It might therefore be expected that CRI rates recorded simultaneously at these positions would be in close agreement. Our results suggest that at least for the population of normal subjects whom we examined, this was not the case. The results of this study and previous published data14 raise doubts as to the robustness of the construct validity for the “core-link” concept as proposed by Sutherland1 and Magoun.3
ICC value
Strength of concordance
<0.4 0.4-0.75 >0.75
Poor Fair to good Excellent
extent to which 2 measurements yield the same result. This is known as concordance. 22 The ICC provides such an index and is the extent of agreement of one measurement on a subject and another measurement on the same subject.21 One assumption that must be met for valid use of ICCs is that of subject heterogeneity. A reduced level of variation among samples can contribute to artificially low reliability coefficients calculated through use of ICCs.27,28 In our study, we assumed that the sample population would present with a range of different CRI rates. The presence of differences between subjects (F = 7.65, df = 10, P = .000) suggests that the assumption of subject heterogeneity was met. A second assumption for valid use of ICCs is that of sample population size.29 The number of subjects (n) should be large enough that n/(n – 1) is close to 1. Our study meets this assumption because n/(n – 1) = 1.1,
Journal of Manipulative and Physiological Therapeutics Volume 24 • Number 3 • March/April 2001 Cranial Rhythmic Impulse • Moran and Gibbons
Although the low levels of interexaminer agreement raise doubts as to the construct validity of the theoretical “corelink” model, the reliability data should also be interpreted in a more clinical context. Di Fabio31 writes that the study of reliability should be accompanied by an explanation of why the measure is important and what influence the measurement has on determining the nature and scope of treatment. One question arising from the interexaminer data is this: Would the palpation of different CRI rates lead the examiners to different diagnostic conclusions and therefore raise the possibility of different and possibly even opposite interventions? Although we did not design our experiment to specifically answer such a question, an indication might be possible on the basis of inspection of the rate plots (Figures 1-4) and consideration of the calculated interexaminer correlation data (Table 1). There were low to nonexistent levels of correlation between examiners palpating simultaneously at the head and the feet (and vice versa) and between examiners consecutively palpating at the same position (nonsignificant r coefficients ranging from –0.314 to +0.45). Inspection of the data for subject 7 in Figure 3 reveals an occasion when one of the examiners palpated a CRI rate (2.8 cycles.min–1) that some proponents would consider to be low or even to be associated with significant pathosis.5,32,33 The second examiner palpating the same subject within minutes of the first recorded a CRI rate (6.07 cycles.min–1) that some proponents would consider to be within the “normal” range.2 In a clinical environment, such discrepancy might have led examiner A to a clinical impression different from that reached by examiner B. In contrast, inspection of Figures 3 and 4 reveals occasions when the examiners were in close agreement with respect to CRI rate. At the head the examiners were in close agreement as to CRI rates for subjects 4, 8, and 9 (Figure 3), whereas at the sacrum the examiners agreed within 1 cycles.min–1 for subjects 4, 5, 6, 8, 9, and 10 (Figure 4). The mean absolute differences between CRI rates recorded by the 2 examiners at various positions are all less than 1.4 cycles.min–1 (Table 2), and though these differences are statistically significant, the clinical importance of the results are less clear. Because a diagnosis of dysfunction is unlikely to be based on assessment of CRI rate alone, reliability studies in the cranial arena should be extended to include other clinically important diagnostic parameters.
CRI Rate Comparisons The mean CRI rates as palpated by the 2 examiners at the head and the sacrum ranged from 2.92 (SD = 0.49) to 4.17 (SD = 0.80) cycles.min–1 (Table 1). The highest single rate recorded was 6.07 cycles.min–1 and the lowest was 2.17 cycles.min–1. These rates are comparable to those recorded in reliability studies.14,20 Wirth-Pattullo and Hayes19 reported slightly faster CRI rates in their reliability study, the mean rates ranging from 4.5 to 7 cycles.min–1. There is little consensus in the literature as to what CRI rate should be considered “normal.”34 Various authors report a range of CRI rates—from 0.6 cycles.min–1 (the slowest) to 10 to 14 cycles.min–1.11,35
That the CRI is generated as a harmonic frequency derived from “multiple biological oscillators,”36 or as a complex interaction of tissue fluid pressure dynamics involving examiner and subject37 is an interesting hypothesis for consideration. Recently, a new hypothesis for the origin of the CRI has been proposed in which the supposed CRI might be felt as an expression of venous vessel wall pulsation, or “venomotion.”38 If future experimental research were to provide evidence for models of CRI generation based on the complex interaction of multiple frequencies (cardiac, respiratory, venous, and so forth) originating from examiner and/or subject, then comparisons of CRI rate between different examiners might become meaningless. In any case, such a finding does not preclude the pressing need for investigation into clinically measurable patient outcomes after treatment through use of cranial techniques. It should be noted that the sample size in this study was small, and some caution should be used in extrapolating the results to the population of cranial practitioners in general. Future studies would benefit from the use of a larger stratified sample of cranial practitioners that would more accurately reflect the age and experience of the cranial practitioner population.
CONCLUSION This study aimed to (1) estimate coefficients of intraexaminer and interexaminer reliability for examiners palpating what they recognized as the CRI, (2) investigate the construct validity of the “core-link” hypothesis, and (3) compare the mean CRI rates recorded by 2 examiners with those reported in the previous literature. Intrarater reliability for the examiners at either the head or the sacrum was fair to good, significant ICCs ranging from +0.52 to +0.73. Interexaminer reliability for simultaneous palpation at the head and the sacrum was poor to nonexistent, the ICCs ranging from –0.09 to +0.31. There were significant differences between rates of the CRI palpated simultaneously at the head and the sacrum. These data fail to support the construct validity of the “core-link” hypothesis as traditionally held by proponents of craniosacral therapy and OCF. The mean CRI rates reported in this study range between 2.92 and 4.17 cycles.min–1 and are comparable to rates reported in similar experimental studies. These mean rates might be viewed by some practitioners as lower than “normal.” Future research should focus on establishing the origin of the CRI phenomenon, further evaluate the reliability of cranial examination procedures (including other diagnostic parameters), and investigate measurable clinical outcomes from such interventions.
ACKNOWLEDGMENTS We thank Ian Fairweather of the Faculty of Human Movement, Victoria University, for providing technical support and the custom designed software application used in this study. We also thank the osteopaths who participated as examiners in this study.
189
190
Journal of Manipulative and Physiological Therapeutics Volume 24 • Number 3 • March/April 2001 Cranial Rhythmic Impulse • Moran and Gibbons
REFERENCES 1. Sutherland WG. The cranial bowl. Minnesota: Free Press Company; 1939. 2. Upledger JE, Vredevoogd JD. Craniosacral therapy. Seattle: Eastland Press; 1983. p. 16. 3. Magoun HI. Osteopathy in the cranial field. Kirksville (MO): Journal Printing Company; 1976. p. 18-9. 4. Wales AL, editor. Teachings in the science of osteopathy. Rudra Press; 1990:22,66,165. 5. Woods JM, Woods RH. A physical finding related to psychiatric disorders. J Am Osteopath Assoc 1961:60:988-93. 6. Sackett DL, Haynes RB, Tugwell P. Clinical epidemiology: a basic science for clinical medicine. 2nd ed. Boston: Little, Brown; 1991. p. 19-49. 7. Vincent-Smith B, Gibbons P. Inter-examiner and intra-examiner reliability of the standing flexion test. Manual Ther 1999; 4:87-93. 8. Bogduk N. An interview with Nicolai Bogduk: musculoskeletal medicine, evidence based medicine and the meaning of life. Newsletter of the Australian Association of Musculoskeletal Medicine 1998. p. 1-4. 9. Leboeuf-Yde C. How real is the subluxation? a research perspective. J Manipulative Physiol Ther 1998;21:492-4. 10. Lay EM. Cranial field. In: Ward RC, editor. Foundations of osteopathic medicine. Baltimore: Williams and Wilkins; 1997. p. 905. 11. Chaitow L. Cranial manipulation theory and practice. Edinburgh: Churchill Livingstone; 1999. p. 22,27. 12. Ettlinger H, Gintis B. Cranial concepts. In: Di Giovanna EL, Schiowitz S, editors. An osteopathic approach to diagnosis and treatment. 2nd ed. Philadelphia: Lippincott; 1991. p. 405. 13. Zanakis MF, DiMeo J, Madonna BFA, Morgan BA, Drasby BA. Objective measurement of the CRI with manipulation and palpation of the sacrum [abstract]. J Am Osteopath Assoc 1996;96:551. 14. Rogers JS, Witt PL, Gross MT, Hacke JD, Genova PA. Simultaneous palpation of the craniosacral rate at the head and feet: intrarater and interrater reliability and rate comparisons. Phys Ther 1998;78:1175-85. 15. Norton JA. Challenge to the concept of craniosacral interaction. Acad Applied Osteopathy J 1996;6:15-21. 16. Chaitow L. Cranial manipulation re-examined. Intern J Alternative Complementary Med 1997. p. 29-32. 17. Butler D. Mobilization of the nervous system. Edinburgh: Churchill Livingstone; 1991. 18. Upledger JE. The reproducibility of craniosacral examination findings: a statistical analysis. J Am Osteopath Assoc 1977; 76:890-9.
19. Wirth-Pattullo V, Hayes KW. Interrater reliability of craniosacral rate measurements and their relationship with subjects’ and examiners’ heart and respiratory rate measurements. Phys Ther 1994. p. 908-20. 20. Hanten WP, Dawson DD, Iwata M, Seiden M, Whitten FG, Zink T. Craniosacral rhythm: reliability and relationships with cardiac and respiratory rates. J Orthop Sports Phys Ther 1998;27:213-8. 21. Shrout PE, Fleiss JL. Intraclass correlations: uses in assesing rater reliability. Psychol Bull 1979;86:420-8. 22. Kramer MS, Feinstein AR. Clinical biostatistics: the biostatistics of concordance. Clin Pharmacol Ther 1981;29:111-23. 23. Fleiss JL. The design and analysis of clinical experiments. New York: John Wiley and Sons; 1986. p. 7. 24. Bartko JJ. On various intraclass correlation reliability coefficients. Psychol Bull 1976;83:762-5. 25. Haas M. Statistical methodology for reliability studies. J Manipulative Physiol Ther 1991;14:119-32. 26. Haas M. The reliability of reliability. J Manipulative Physiol Ther 1991;14:199-208. 27. Lahey MA, Downey RG, Saal FE. Intraclass correlations: there’s more than meets the eye. Psychol Bull 1983;93:586-95. 28. Mitchell SK. Interobserver agreement, reliability, and generalizability of data collected in observational studies. Psychol Bull 1979;86:376-90. 29. Fleiss JL. Measuring agreement between two judges on the presence or absence of a trait. Biometrics 1975;31:651-9. 30. Sibley G, Broder-Oldach RE, Norton JM. Interexaminer agreement in the characterization of the cranial rhythmic impulse [abstract]. J Am Osteopath Assoc 1992;92:1285. 31. Di Fabio RP. Studies of reliability: show me the question [editorial]. J Orthop Sports Phys Ther 1999;29:370-1. 32. Karni Z, Upledger JE, Mirzahi J, Heller L, Becker E, Najenson T. Examination of the cranial rhythm in long standing coma and chronic neurological cases. In: Upledger JE, Vredevoogd JD. Craniosacral therapy. Seattle: Eastland Press; 1983:275-81. 33. Brookes D. Lectures on cranial osteopathy. Wellingborough, United Kingdom: Thorsons Publishers; 1981. p. 56-9. 34. Chaitow L. Cranial dogma re-assessed. Intern J Alternative Complementary Med 1997. p. 28-32. 35. Greenman PE. Principles of manual medicine. 2nd ed. Baltimore: Williams and Wilkins; 1996. p. 165. 36. McPartland JM, Mein EA. Entrainment and the cranial rhythmic impulse. Alternative Therapies 1997;3:40-5. 37. Norton JM. A tissue pressure model for the palpatory perception of the cranial rhythmic impulse. J Am Osteopath Assoc 1991;91:975-94. 38. Farasyn A. New hypothesis for the origin of cranio-sacral motion. J Bodywork Movement Ther 1999;3:229-37.