Effect of positional asymmetry palpatory models on improvement and retention of accuracy during pelvic asymmetry assessments

Methods

The current study utilized a prospective design to determine the effect of training with positional asymmetry models on pelvic asymmetry assessment accuracy over a 3-week duration. All study procedures (examinations and practice times) were completed during June and July of 2022 and 2023. The study was deemed exempt by the A.T. Still University-Kirksville Institutional Review Board, and no funding was required. Five first-year osteopathic medical students (OMS1) and seven undergraduate students (novice) were recruited from those interested in being part of the medical student research elective in Osteopathic Manipulative Medicine or the undergraduate Clinician Researcher Development Program. By enrolling in the research internship or elective, students provided consent to participate in research activities. Participant age and dominant eye were recorded. Examiners were instructed to stand on the opposite side of the table of their eye dominance and to center their eye over the middle of the treatment table when performing the testing.

Two types of positional asymmetry models spatially imitated ASIS or PSIS landmarks (Figure 1). Each model consisted of six pairs of 12.7 × 3.175 mm rectangular metal ridges separated by 24.0 cm on ASIS models and by 11.5 cm on PSIS models. Each ridge was independently adjustable in 1-mm increments (toward the head or foot of the treatment table) to create asymmetries up to 12 mm. A 3-mm thick thermoplastic elastomer sheet was placed over the palpation surface of the models to hide visual cues and simulate soft tissue. Models were secured in place to prevent shifting and to maintain a central position relative to the table. Examiners adjusted the table height as needed.

A 72-question assessment test, consisting of 36 ASIS and 36 PSIS questions, was created and accessed through Research Electronic Data Capture (REDCap) software (Nashville, TN). The examiners had to determine whether the right-side landmark was superior, inferior, or equal to the left. Asymmetry values were set at −8, −6, −4, −2, 0 (equal), 2, 4, 6, or 8 mm. Positive and negative values were assigned to asymmetries to indicate the direction: positive values indicated that the right landmark was superior to the left, and the negative values indicated that the right landmark was inferior to the left. Each asymmetry value appeared four times at random in the assessment. The answers were scored in REDCap.

Initially, examiners observed a 30-min demonstration of the assessment, including time to identify their dominant eye, ask questions, and practice the process. They then completed the baseline assessment. Examiners trained on the models for 5 h per week (1 h/day) over 2 weeks with self-directed practice and objective feedback. Incorrect assessments required participants to remove the gel pad and re-evaluate to identify errors. After 2 weeks of training, there was a 1-week washout period. Accuracy was evaluated at baseline, midpoint (1 week of training), final (2 weeks of training), and retention (1-week post training). To minimize any potential priming effects, half of the examiners began each assessment on the ASIS models, whereas the other half started on the PSIS models.

For all assessments, accuracy was scored as the correct determination of asymmetry. Because of the potential correlation of the assessments made by the same examiner, a mixed-effect model was utilized to investigate the effect of training. By including the examiners as random effects, our mixed-effect model extended a three-way analysis of variance (ANOVA) to accommodate the correlated assessment outcomes. In addition to the random effects of examiners, the following were included in the model as fixed effects: the asymmetry values of the ASIS and PSIS models, weeks (baseline, midpoint, final, and retention), the examiners’ preferred side of the table, and all of the two-way and three-way interactions. The model was fitted based on 1728 ASIS and 1728 PSIS responses. Estimated marginal means and standard errors (SE) were utilized to quantify and test for changes in scores from week to week and overall (baseline to retention). Retention was assessed utilizing estimated marginal means and SE to determine if the side of the table affected overall accuracy. The influence of the examiners’ side of the table on accuracy for specific asymmetry values (2, 4, 6, and 8 mm) in both the superior and inferior outcomes was also evaluated. Analyses were performed utilizing SPSS version 29.0 (IBM Corp., Armonk, NY), and p<0.05 was considered statistically significant.

Results

Twelve students (mean [SD] age, 23.8 [2.9] years; range, 20.3–29.2 years) participated as examiners. Of these, seven students had no prior palpation training, and five had completed a year of osteopathic medical school. Based on their dominant eyes, six students stood on the left side of the treatment table, and six stood on the right.

The estimated mean scores for the ASIS models are presented in Table 1. Scores by examiners’ preferred side of the table are presented in Appendix 1. The estimated marginal mean score at baseline was 57.6 %. Overall, the accuracy scores improved from baseline to midpoint (+18.9 %, p<0.001) and from midpoint to final (+6.6 %, p=0.01). The scores decreased from final to retention (−7.2 %, p=0.01), but the overall scores from baseline to retention improved (+18.3 %, p<0.001). The OMS1 students performed better than the novice students at baseline on the ASIS landmarks (OMS1, 70.6 %; novice, 48.4 %; p=0.003), but no differences were observed at the midpoint (OMS1, 75.2 %; novice, 77.5 %; p=0.73), final (OMS1, 87.2 %; novice, 80.2 %; p=0.29), or retention (OMS1, 80.0 %; novice, 73.0 %; p=0.29) assessments.

Figure 1:

Positional asymmetry models for (A) anterior superior iliac spine (ASIS) landmarks and (B) posterior superior iliac spine (PSIS) landmarks. L, left-side landmark; R, right-side landmark.

The estimated mean scores for the PSIS are presented in Table 1. The scores by examiners’ preferred side of the table are presented in Appendix 1. The estimated marginal mean score at baseline was 72.9 %. Overall, the accuracy scores improved from baseline to midpoint (+13.0 %, p<0.001). Although the scores changed from midpoint to final (−0.7 %) and from final to retention (+2.7 %), they were not significant (both p>0.21). The overall scores from baseline to retention improved (+15.0 %, p<0.001). The OMS1 students performed better than the novice students at baseline on PSIS landmarks (OMS1, 87.4 %; novice, 62.7 %; p<0.001), but no differences were observed at the midpoint (OMS1, 86.2 %; novice, 85.7 %; p=0.91), final (OMS1, 87.8 %; novice, 83.3 %; p=0.37), or retention (OMS1, 90.0 %; novice, 86.5 %; p=0.48) assessments.

Performance differences based on the examiners’ side of the table were compared to baseline results. Overall, ASIS scores did not significantly differ between the left (59.7 %) or right (55.6 %) sides (p=0.55). However, PSIS scores were higher for those on the left side (79.2 %) compared to the right (66.7 %, p=0.02) (Table 2). Examiners on the right side had better ASIS scores for equal (+33.3 %, p=0.003), 2 mm (+54.2 %, p<0.03), and 8 mm (+25.0 %, p<0.03), but worse for all PSIS superior asymmetry values (all p<0.03). They also had worse ASIS scores for all inferior ASIS asymmetry values (all p<0.001), but better scores or the inferior PSIS asymmetry value of −2 mm (+37.5 %, p<0.001).

Comparisons between the examiners’ side of the table and performance for asymmetry outcomes are presented in Table 3 and Appendix 2. For those examining from the left side, ASIS scores for superior asymmetries were lower than inferior asymmetries at baseline (−2 mm, −50.0 %; −6 mm,−25.0 %; −8 mm,−20.8 %; all p<0.04), at midpoint (−4 mm,−20.9 %; p=0.04), at final (−2 mm, −33.3 %; p<0.001), and at retention (−2 mm, −20.8 %; −4 mm, −20.8 %; both p=0.04). Those on the right side of the table had higher ASIS scores for superior asymmetries at baseline (all p<0.001), midpoint (2 mm, +58.3 %; 4 mm, +41.6 %; both p<0.001), final (2 mm, +25.0 %, p=0.01), and retention (2 mm, +29.1 %, p=0.004).

For those examining from the left side, PSIS scores for superior asymmetries were higher than inferior asymmetries at baseline (2 mm, +62.5 %; 4 mm, +33.3 %; both p<0.001) and final (2 mm, +25.0 %; p=0.01). Those on the right side had lower PSIS scores for superior asymmetries at baseline (−2 mm, −20.8 %; −4 mm, −45.8 %; −6 mm, −20.8 %; all p<0.04), midpoint (2 mm, −37.5 %, p<0.001); and final (2 mm, −45.9 %; p<0.001).

Discussion

In the current study, we investigated the effect of training with positional asymmetry models on the improvement and retention of accuracy with pelvic landmark assessments for asymmetry. At baseline, accuracy was better for asymmetries on PSIS than ASIS landmarks (Table 1). Subsequent assessments indicated improvement in the overall scores for both landmarks at midpoint and for the ASIS landmarks at final (Table 2). Retention assessment showed sustained improvements for PSIS models but a decline for ASIS models even though both remained improved from baseline. Our findings suggested that objective feedback with quantified asymmetries significantly enhanced the accuracy of identifying positional asymmetries on simulated models.

Results also indicated that larger asymmetry values and spatially closer landmarks were more accurately identified (Table 1). The ASIS assessments, where the landmarks are over 2 times farther apart than the PSIS assessments, appeared to be more prone to errors. It seems that the greater the eye movement, the more difficult it is to accurately assess the landmarks.

With training, improvements began to plateau for both landmarks at the midpoint assessment (especially for asymmetries ≥4 mm), probably caused by a ceiling effect. The accuracy of the ASIS assessments were usually at least 10 % lower than the accuracy for the PSIS assessments. Additionally, although students with 1 year of palpatory experience had greater accuracy than those with no experience at baseline, there were no statistical differences in accuracy between the groups on the midpoint, final, or retention assessments for both models. This result suggested that, within 5 h of practice, the skills of students with no experience were comparable to those with 1 year of standard curricular training. However, our results also showed consistently higher, but not statistically significant, performance trends for the OMS1 students at the midpoint, final, or retention assessments. It is possible that the current statistical findings may not be telling the whole story because of the small sample size. Additional studies with larger sample sizes are necessary to evaluate actual performance differences between the two groups. This line of research is necessary in establishing evidence-based curricular designs.

The results of the current study had similarities to previous research. Like the study by Stovall et al. [18] we evaluated asymmetry at 2-mm increments in both directions of asymmetry, which allowed for a higher level of precision compared to the study by Lee et al. [19], who utilized 5-mm increments. Unlike previous studies, we tested each asymmetry four times, twice for each direction of asymmetry, which minimized the statistical influence of chance correctness. We also specifically evaluated performance patterns for those standing at different sides of the table, focusing on accuracy characteristics vs. reliability as previously reported [18]. We found that those standing on the left side of the table (right eye dominant) more consistently saw the right ASIS inferior and the right PSIS superior. For those examining from the right side of the table (left eye dominant), it was the opposite. They more consistently saw the right ASIS superior and the right PSIS inferior.

It was unexpected to see that the examination location would show better accuracy with opposite findings. Because of the design of the models, the introduction given during study orientation demonstrating how to localize the landmarks, and the consistent trends seen in the data, it seemed unlikely that poor landmark localization was the cause for the data pattern. Although we cannot rule out the visual system as a potential source for error, it seemed that visual perception would be constant during the study period. Looking more closely at the data, it seems that accuracy was consistently better when the landmark on the opposite side of the table was superior to the landmark on the same side of the table as the examiner (Table 2 and 3, and Appendix 1 and 2). Looking at the students’ examination process, it appeared that student examiners were setting up an internal frame of reference based on their head position and posture (Figure 2). Comparing eye and body position with the data supported a predictable skewing of zero and consistently poor accuracy in one direction of asymmetry. If the dominant eye is not in the midline between the landmarks and the eyes are not in the same horizontal plane as the landmarks being assessed, the examiner’s internal frame of reference is skewed, and a region associated with smaller asymmetries (Figure 2 – pink shaded zone) will appear correct to the examiner but will actually be incorrect. With practice, adjusting posture so the dominant eye was in the midline of the model and the shoulders and head were in the horizontal plane of the table and model, the curves seen in Appendix 2 shifted to have the lowest point at zero, the shape of the curves looked more symmetrical for both positive and negative asymmetries, and the overall score demonstrated improved accuracy.

Figure 2:

Impact of examiner body position on landmark localization. (a) Actual horizontal plane; (b) perceived horizontal plane; (c) ideally positioned examiner dominant eye/line of view; (d) misaligned dominant eye; (e) towards the head; (f) towards the feet; (g) shaded area-error zone.

From an examiner’s perspective, the poor body position during palpatory assessment seemed to be an underrotation or relaxation of the trunk, shoulders, and head. Through training with the palpatory models, the student examiners identified ways to mitigate these errors by lowering the height of the table in order to establish a more comfortable examination posture aligned with the cardinal planes of the model. This learned behavior also seemed to reduce the skewed side-specific misidentification of asymmetry (Appendix 2).

The independent, time-based training utilized in this study (1 h/day, 5 days/week for 2 weeks) allowed us to quantify the amount of time necessary for our student examiners to improve and sustain their skills. Further, the flexibility of this training approach allowed the examiners to incorporate self-assessment into their personal schedules and to make evidence-informed individualized postural adjustments to their skills. Consequently, this training approach does not require additional formal laboratory time to achieve improved performance.

The primary limitation of the current study was that only 12 OMS1 and novice examiners participated, so the results may not be generalizable to larger groups or to learners with more palpatory experience. Additionally, our positional asymmetry training models have identical features, so they do not account for human landmark anatomical variability. The models lack the variability in soft-tissue and landmark characteristics typically observed in humans. Although the models were covered with a thermoplastic elastomer sheet to simulate soft tissue, it cannot fully replicate human tissue. Additionally, centering humans on a treatment table is more challenging than positioning models, and this difference can complicate an examiner’s determination of an ideal frame of reference.

Currently, palpatory training with humans lacks objective feedback. As such, these models were specifically designed to be simplified and idealized human approximations to help student examiners utilize objective feedback to develop awareness of basic examination principles and reinforce ideal performance habits. Ideally, this objective feedback during training may increase test accuracy on humans. At this time, we believe that the use of models provides a method for developing essential motor skills that can be applied when examining humans, which may potentially improve the reliability of such findings. However, it is still unknown how the palpatory skills developed through training with the models will translate to examining human patients.

Future investigations should include a larger sample size, such as an entire medical school class. Studies should also evaluate the performance accuracy of practicing osteopathic physicians to determine the effects of time-based training for those with expert-level experience. Other studies, similar to the work of Lavazza et al. [21] should consider investigating how the frequency and duration of training sessions affect palpatory accuracy. Similarly, future studies should explore the impact of extended training gaps on accuracy and evaluate the optimal frequency and duration of recalibration. Ultimately, this line of research aims to objectively characterize palpatory skills for assessing pelvic landmark asymmetry and establish the clinical relevance of this test.

Conclusions

The current study found that the ability of student examiners to accurately identify asymmetries decreased if the landmark on their side of the treatment table was superior. However, training with positional asymmetry models and objective feedback significantly improved palpatory accuracy for both ASIS and PSIS landmarks after 5 h over 1 week. Continued training further enhanced accuracy for the ASIS landmarks and maintained it for PSIS landmarks. These results underscored the value of objective feedback for developing palpatory skills by helping examiners accurately assess challenging asymmetries. Incorporating positional asymmetry models into medical training appears to be a promising educational tool for standardizing and improving pelvic asymmetry assessments.