For the results of a specific palpatory test to provide useful information that can be communicated between clinicians, that test must have acceptable interexaminer reliability. However, although such reliability is important, it does not necessarily define the clinical usefulness of a test or method of assessment.¹⁹ For instance, although cardiac auscultation has not been clearly demonstrated to have adequate interexaminer reliability,^20,21 it is still used in clinical practice along with history taking, physical examination, and other diagnostic testing to assess clinical presentation in a patient encounter.¹⁹ As pointed out by Joshua et al,²² there is a need for improved understanding in many forms of clinical examination.²² From this perspective, palpation for bony anatomic landmark asymmetry is similar to cardiac auscultation, highlighting the need for critical analysis of methods used in clinical examination. 

Most methods of palpatory examination in manipulative medicine lack a reference standard, which further increases the importance of reliability.¹⁹ In this context, if two examiners can be demonstrated experimentally to consistently agree about diagnostic findings, there is a strong likelihood that what is being tested for exists.¹⁹ Therefore, demonstrating the reliability of palpatory methods in manipulative medicine is of primary importance. 

A variety of statistical methods were used in early reliability experiments in manipulative medicine.^19,23,24 The current standard of reliability analysis is the κ statistic, which has gained widespread acceptance as a measure of reliability because it accounts for the role of chance in dichotomous agreement.^23,24 As pointed out in a recent literature review²⁵ of manual examination in the spine and updated reliability criteria, there are limitations to the use of this statistic.^15,25 Prevalence index and selection of patient population are two important elements that must be carefully addressed when making a qualitative interpretation of κ values, because they can lead to falsely low or high κ values.¹⁵ 

Two methods currently exist to assess the quality of reliability studies in manipulative medicine.^15,25 Stochkendahl et al²⁵ devised a 6-point system based on clinical research guidelines to assess the method quality of reliability studies. These criteria were used in our review and can be found in Figure 1. More recently, the Scientific Committee of the International Academy for Manual/Musculoskeletal Medicine (IAMMM)¹⁵ developed a 78-point system for reviewing reliability studies. This protocol evaluates four primary domains in analysis of reliability studies: study design, study population, diagnostic procedures, and data analysis and presentation.¹⁵ We did not use this protocol in the present study because it does not clearly define analysis of asymmetry assessment. Although to our knowledge this protocol has not yet been used in literature analysis, it provides a useful guide for developing and assessing future reliability studies. 

Attempts to experimentally evaluate the reliability of musculoskeletal assessment have explored various assessment methods, such as motion testing and pain provocation testing of the sacroiliac joint. For motion testing and pain provocation testing, current literature reviews exist.^26-29 However, research into assessment of anatomic landmark asymmetry as defined in manipulative medicine is relatively new, and investigation into this area has evolved substantially in the past 10 years. Our initial literature search identified two systematic literature reviews^25,26 assessing the reliability of a variety of palpatory methods, including static palpation. While completing the present review, we identified a comprehensive literature review specific to assessment of static anatomic landmark asymmetry,³⁰ which provides a broad overview of research investigating the reliability of static segmental identification and assessment of bilateral anatomic landmark asymmetry in the entire spine and pelvis. Our review incorporates two more recent studies and serves as a more focused complement to this recently published review. Thus, our goal was to analyze and assess the quality of current research investigating the reliability of assessments for bony anatomic landmark asymmetry in the lumbar spine and pelvis. We also sought to discuss these findings in the context of osteopathic medical education. 

We searched the MEDLINE, CINAHL, AMED, MANTIS, Academic Search Complete, and Web of Knowledge literature databases by using multiple combinations of the following keywords: palpation, asymmetry, inter or intraexaminer reliability, tissue texture, assessment, and anatomic landmark. Inclusion criteria were as follows: studies that assessed reliability of palpatory tests for static asymmetry of L1 though L5 transverse processes or the pelvic or sacral anatomic landmarks, experimental studies in which the design was appropriate to answer the raised question(s), and studies published in peer-reviewed scientific journals. 

Excluded were studies that did not experimentally evaluate the reliability of assessments for bony anatomic landmark asymmetry, evaluated only the reliability of segmental identification of spinal levels, evaluated only iliac crest asymmetry assessment, or provided opinions without an experimental framework. 

The initial search identified 40 articles of interest. Initial screening for subject relevance reduced the list to 18 articles. Of that group, 6 articles did not meet inclusion criteria because of factors identified above. Reviewing the references of all articles after the initial screen revealed 5 further articles of interest that did not meet inclusion criteria. Thus, 23 studies were identified, 9 of which met inclusion criteria (Table). 

The posterior superior iliac spine (PSIS), inferior lateral angles (ILAs), and sacral sulcus (SS) were the posterior pelvic landmarks assessed in the reviewed studies. The ILAs are assessed for asymmetry in the transverse (ILA-A/P [anterior/posterior]) and coronal (ILA-S/I [superior/inferior]) planes. Some authors differentiate the SS and the sacral base in palpatory assessment, but it is our opinion that there is no clinically palpable difference between the landmarks.^1,2,40,41 For the purpose of this review, “SS” will be used to describe the sacral base, as in some of the reviewed articles. The SS is evaluated by assessing depth medial to the PSIS through palpation rather than visual assessment. This “landmark” was included because historically it has been assumed to give information regarding sacral position. This assumption has since been challenged, and the finding seems to correspond more directly to the multifidus or soft-tissue density than to sacral position.⁴¹ 

The only landmark evaluated in the supine position was the anterior superior iliac spine (ASIS). In muscle energy technique, the ASIS is assessed for asymmetry in the transverse, coronal, and sagittal planes.^4,42,43 Only research examining ASIS symmetry in the coronal plane (superior/inferior displacement) was found.^31,33,34-39 Lumbar transverse processes were evaluated by palpation for asymmetry in various planes and in variations of flexion, extension, and neutral position.^35,36 

The 6-point quality assessment scale was implemented in the analysis (Figure 1). The first author performed assessments, and the second author reviewed the analysis. The modified 4-point intraexaminer reliability scores were not included in the Table, because the emphasis of this study was to assess the overall quality of reliability studies. 

Results of quality analysis using the 6-point scoring system ranged from 2 (33%) to 6 (100%). The 6-point scores are shown with κ values in the Table. The ranges for κ statistics without training for interexaminer reliability were as follows: ASIS, -0.01 to 0.19; PSIS, 0.04 to 0.15; ILA-A/P, -0.03 to 0.11; ILA-S/I, -0.01 to 0.08; SS, -0.4 to 0.37; and lumbar spine transverse processes L1 through L5, 0.04 to 0.17. Ranges for intraexaminer reliability were higher for all associated landmarks: ASIS, 0.19 to 0.4; PSIS, 0.13 to 0.49; ILA-A/P, 0.1 to 0.2; ILA-S/I, 0.03 to 0.21; SS, 0.24 to 0.28; and lumbar spine transverse processes L1 through L5, not applicable (none of the reviewed studies reported intraexaminer reliability of lumbar spine transverse process asymmetry assessment). Results are summarized in the Table. 

Criteria for reliability studies of clinical tests suggest using both symptomatic and asymptomatic patient populations.^44,45 The goal is to provide a more accurate spread of the data in question. This is thought to be important, because the prevalence of positive findings is assumed to be higher in a symptomatic patient population. However, as Kmita and Lucas³¹ pointed out, the prevalence of bony anatomic landmark asymmetry in these populations has not been defined. To address the need for a balanced patient population, those authors selected a population that included symptomatic and asymptomatic patients with low back pain. To our knowledge, this was the only study identified that evaluated a mixed population. Also supporting the use of symptomatic and asymptomatic populations was the assumption that examiners could be biased if they were aware that the population was only symptomatic or only asymptomatic. Other experiments using only symptomatic patient populations, however, have not demonstrated increased reliability for clinical tests of static asymmetry.^32,33,39 

In addition to subject selection criteria, another important component of reliability studies is inclusion of an adequate number of subjects and examiners. Current guidelines suggest that the optimum number of subjects for reliability experiments is 25 to 40 subjects and that the optimum number of examiners is 2. The reviewed experiments demonstrated a wide range in numbers of examiners and subjects; the number of subjects ranged from 9 to 77, and the number of examiners from 2 to 10. Only two studies^32,38 met the criteria for subject number, and three studies^32,33,38 met the criteria for examiner number. All reviewed experiments used examiners with substantial training and varying levels of clinical experience (Table). Given the variation in subject and examiner number in the reviewed experiments, the ability to compare studies directly is limited. 

The number of tests evaluated in a reliability study is also important in study design. Recently updated guidelines recommend studying only one test at a time for experiments designed to assess the reliability of a specific test. O'Haire and Gibbons³⁷ performed an experiment in which posterior pelvic and sacral landmarks were evaluated by 10 fifth-year Australian osteopathy students. One of the challenges noted was the 1200 examinations required of each examiner, which increased the chance for examiner fatigue. This challenge, coupled with the amount of time subjects were required to lie prone on the examination table, was a major experimental confounder. In the clinical setting, asymmetry assessment in the pelvis requires multiple assessments to determine the side of dysfunction and final diagnosis.^2-4,7,9 Although a single assessment of asymmetry is not typical of clinical models of asymmetry assessment, only the study by Spring et al³⁶ and a study of medial malleolus asymmetry⁴⁶ have investigated isolated evaluation of one landmark for asymmetry. 

Along with limiting the number of tests and having adequate numbers of subjects and examiners, the calibration of specific techniques is another important element of reliability research. Two of the reviewed studies predetermined what constitutes a positive finding.^31,34 The experiment by Fryer et al³⁴ initially had the examiners reach an agreement on how much asymmetry constituted a positive finding, “approximately” 3 mm. The other experiment set 3 mm as a positive finding.³¹ Implementing such a threshold, however, introduces another confounder, because the method of doing so was not adequately described or discussed in any of the studies. There is no known ability to accurately determine amounts of asymmetry in vivo, and it is unlikely that the standardization sessions provided a mechanism for substantially improving the reliability of assessments. This finding was further confirmed by the results of these two experiments. 

For the evaluation of low κ values in reliability research, the prevalence index has been suggested to be of primary importance.¹⁵ The prevalence index is defined as the frequency of positive judged tests in the study population.¹⁵ In applying the prevalence index in reliability analysis, only one palpatory test with dichotomous findings can be used.¹⁵ In assessments for asymmetry of bony anatomic landmarks in the pelvis, however, three options are often given to examiners: right side greater than left side, both sides equal, and left side greater than right side. Six studies offered three options in assessment.^31-34,37,39 Examiners were given two to six options for assessing the lumbar spine but only one to evaluate for dichotomous asymmetry in the pelvis.^35,36,38 Thus, the role of the prevalence index in the assessment of asymmetry is unclear. When the prevalence index is incorporated into studies, a calibration session is needed to attain more than 85% agreement before conducting the experiment. This allows the prevalence index to be within optimum range to interpret the κ statistic. Therefore, qualitative comparison of κ values is limited for studies that do not provide information about the prevalence index of a given test. Readers are referred to the IAMMM guidelines for reliability studies for further discussion of the prevalence index.¹⁵ The prevalence of findings was determined in two studies, but no further mention of associated data could be found in these studies. The absence of this measure in reliability research highlights an important area for improvement. 

Two experiments examined the role of formal training programs with the goal of improving reliability.^34,35 Two methods of training were implemented. One experiment used two groups of five examiners, with one group receiving training over two 1-hour training sessions with an experienced clinician.³⁴ During these training sessions, technique was standardized, and examiners evaluated and compared assessments. Interestingly, examiners demonstrated substantial improvements in reliability after training for some landmarks but not for others.³⁴ Assessments of the ASIS have consistently had higher reliability than assessments of other landmarks in the pelvic girdle, and they improved after training.³⁴ This improvement was not enough to reach statistical significance, but the change suggests that training changed something about how asymmetry was assessed in the trained group. 

Another form of training introduced by Degenhardt et al³⁵ consisted of “consensus training,” which was focused on training examiners to perform assessments in as similar a manner as possible. This method of training was developed to better understand the known low interexaminer reliability of palpatory assessment. When results differed between examiners, the researchers compared evaluation techniques and attempted to understand the reason for observed differences.³⁵ The results of this experiment demonstrated statistically significant increases in interexaminer reliability after this method of training. 

These results, however, must be interpreted with caution because of methodologic inconsistencies and nonstandardized changes. In the final two phases of the experiment, patients did not change positions during assessment until all examiners had evaluated them. This was different from phase 1 in that all assessments, which included active and passive motion testing, were conducted before another examiner made assessments. Although these limitations affect interpretation, having examiners with a high consensus as to what constitutes a positive finding is probably significant in musculoskeletal palpatory assessment. 

In recent literature reviews of palpatory methods in manipulative medicine, intraexaminer reliability has consistently been higher than interexaminer reliability.^25,26,28-30 In the studies we reviewed, this remained true except in one experiment.³⁶ Examiners intuitively can be expected to agree with themselves more than with other examiners, but the reasons for this commonly observed phenomenon are not well understood. The accuracy and consistency of finger placement as well as visual cues have been suggested to play a role in low interexaminer reliability.^34,37 Although standardized training sessions were used in two experiments, the accuracy of finger placement for individual examiners was not specifically addressed.^31,34 

To date, minimal research has been conducted into methods of bilateral asymmetry assessment in manipulative medicine. The varied methods in the literature limit the comparison of reliability studies. The quality assessment method that was used did not take prevalence of findings into account. Our review seeks to clarify the role of static asymmetry assessment of lumbar spine and pelvic landmarks, but owing to the varying sizes of these landmarks and differences in methods, the ability to make direct comparisons will remain limited until future studies explore this method of asymmetry. 

Not all osteopathic medical students begin their education with a desire solely to practice manipulative medicine. However, most students approach their initial osteopathic manipulative medicine (OMM) courses with a genuine intellectual curiosity about OMM techniques. These students learn about Fryette's laws, pelvic and sacral mechanics, and concepts of palpatory diagnosis and treatment throughout their first 2 years of education. These concepts are often taught as basic facts with little discussion of validity or reliability, but there is growing evidence that some concepts are invalid or at least incompletely understood.^47-52 This fact marks one of the single greatest challenges facing osteopathic medical education today: how do educators best teach traditional, clinical models of palpatory diagnosis that the growing body of evidence suggests to be unreliable or invalid? 

As previously discussed by Fryer,⁴⁷ various approaches can be used to address this issue in the OMT lecture hall or laboratory; one can ignore the challenge, hoping it is not brought up by inquisitive students, or address it directly by incorporating the role of critical thinking into the OMT curriculum.⁴⁷ It is our opinion that allowing students to develop hands-on skills from highly trained instructors while at the same time thinking critically of concepts presented will prove beneficial for the profession. Our review and previous analysis of palpatory methods can serve as a starting point for critical discussion of the palpatory methods found in manipulative medical professions.^19,47,51 This approach can help cultivate respect for the history of diagnostic approaches in osteopathic medicine while allowing growth and adaptation of current methods and student skills. Meeting the challenges of teaching today's osteopathic medical student with explanations that address current scientific knowledge in the most uniquely osteopathic component of osteopathic medical education can only foster further understanding of and interest in the art and science of osteopathic medicine. Research continues to grow and establish itself in the osteopathic profession as never before, and we in the profession have the opportunity to equip future osteopathic physicians with both a high level of manipulative skill and an appreciation for the balance of clinical and evidence-based knowledge. 

Overall improvement in the number of high-quality reliability studies with similar methods is paramount. With the increasing need for an evidence base in medical professions, the objective understanding of palpatory assessment methods is especially important for those professions with an emphasis on manipulative medicine. When evaluating a specific test, studies should follow the IAMMM protocol to allow standardization of experimental methods. Furthermore, the manner in which expert practitioners of manipulative medicine developed a high degree of diagnostic skill must begin to be experimentally established. Another emerging goal is to establish detection thresholds for pelvic landmark positional assessment to aid in training.⁵³ The use of highly controllable models should be explored to understand the degree of accuracy in human perception. The extent of bony anatomic landmark asymmetry that may be normal and asymptomatic needs to be quantified. Through such advancements in knowledge, the profession can strengthen the scientific base needed to better explain the diagnosis of somatic dysfunction. 

The key findings of the present review article are presented in Figure 2. The quality and methods of studies investigating the reliability of assessments for bony anatomic landmark asymmetry are variable. Further research is needed to understand the reliability of evaluation methods in manipulative medicine. Assessment of bony anatomic landmark asymmetry is only one component of the diagnostic paradigm in manipulative medicine, but an improved understanding of its reliability would be of substantial value to the osteopathic medical profession, as well as other professions in manipulative medicine.