Evaluation of an Ultrasound-Guided Palpation Intervention for Palpation Skill Training: A Single-Blind, Parallel Group, Randomised Controlled Trial


 Although there are increasing reports on the usefulness of sonopalpation with ultrasound imaging, many previous studies have reported interventions without a control group. This single-blind, parallel group randomised controlled trial aimed to determine whether educational instruction with sonopalpation for medical students has a superior effect on skill improvement than traditional instruction without ultrasonography. Twenty-nine medical students participated in the study and were randomised using block randomisation into an ultrasound imaging group (n = 15) and a control group (n = 14). Then, they underwent three training sessions focusing on the shoulder joint. Participants underwent a scoring assessment of their palpation skills at pre-intervention, post-intervention, and follow-up 3 months after training. The raters were blinded to the subjects’ group. The Friedman and Mann–Whitney U tests were used for data analysis. The intervention group showed a significant increase in scores at post-intervention and at the 3-month follow-up when compared with the pre-intervention group (p < 0.05); the effect sizes were large (0.849 and 0.849, respectively). A comparison of the scores at different time points after the intervention at the 3-month follow-up revealed that there was no significant difference between the groups. Education using ultrasound images may improve the sonopalpation skills of medical students.


Introduction
Obtaining accurate physical examination ndings is one of the most basic and important skills for medical doctors and physical therapists. However, over the past 10 years, reports on palpation skills of medical doctors, physical therapists, and medical students have shown that the accuracy is generally low for each joint [1][2][3] . Accurate palpation of anatomical landmarks is imperative for an accurate diagnosis and/or judgement of the pathology or cause of pain 4 .
To address this problem, there have been increasing reports in recent years on the usefulness of ultrasound (US) imaging for palpation of musculoskeletal tissues 5,6 . Sonopalpation, a palpation technique that uses US imaging systems, is useful for diagnosing and identifying the cause of pain and improving the accuracy of diagnosis 7,8 . US imaging allows non-invasive and real-time evaluation compared with other diagnostic imaging techniques. It is used to identify anatomical landmarks in many departments, including the physical medicine and rehabilitation areas 9 . Additionally, systematic learning courses for improving palpation skills using US images have been developed to take advantage of the US images recently 10 . There has also been an increasing number of interventional studies on the usefulness of US imaging as a tool to improve the palpation skills of medical students or residents [11][12][13][14][15] .
Among the many joints in the human body, the shoulder joint, in particular, is important for residents and physical therapists to palpate accurately to understand the cause of pain, dysfunction, and potential pathological causality 4 . It is estimated that approximately 20% of the population experience shoulder pain 16 . For the management of such patients, proper identi cation and classi cation of the cause of pain are important. The ability to perform accurate palpation allows appropriate clinical assessment and clinical reasoning of pathological conditions, such as injuries and abnormalities in the tissues surrounding the shoulder joint 17,18 . However, the insu cient palpation skills of medical students and inexperienced clinicians pose a problem.
Many preliminary studies have reported improvements in palpation skills using US imaging. However, most of these were observational studies that did not examine confounders or interventional studies without a control group. They also had various limitations, such as insu cient blinding in an intervention trial. The results of these studies did not provide su cient evidence for the e cacy of sonopalpation. In the present study, we examined whether a palpation lecture using US images for medical students is more useful in improving their skills than a traditional palpation lecture. We hypothesised that a course that uses US imaging ndings will improve medical students' shoulder joint palpation skills more than a course that focuses on the conventional method. We focused on the shoulder joint tissue because it is clinically important to palpate this joint tissue accurately and sought to provide higher-quality evidence by conducting a single-blind randomised controlled trial.

Study design and participants
The study protocol conformed to the Declaration of Helsinki. The research was conducted with approval from the Medical Research Ethics Committee at Shinshu university (no. 4372), and our research protocol was registered with UMIN-CTR (trial number: UMIN000038286) on 15 October 2019.
Twenty-nine rst-year college students from a single university, gathered through publicity via campus bulletin boards and other means, participated in a hands-on training course on palpation during 2019-2020. After the subjects were fully informed about the details regarding participation in this study, their written informed consent was obtained. The exclusion criteria for participants were as follows: 1) those with sensory impairment in the ngers; 2) medical students, those who had worked in the medical or health-related eld in the past, or those who had taken a course to acquire palpation skills in the past; and 3) any other person who was considered inappropriate as a subject by the principal investigator for valid reasons. Data collection and implementation were performed in the practical training room of the university. Of the 38 participants who received explanations, 29 who consented to participate in the study were randomly divided into two groups: the US imaging-based sonopalpation education group (US group) and the conventional palpation education group (control group). Randomisation was performed using a block randomisation method with a block size of 4. The order of allocation was concealed until each group was allocated. The order of assignment was made by those who were not intervention implementers or evaluators.

Methods
For each group of participants, before the course started, a scoring assessment of the participants' palpation skills in terms of accuracy and time was conducted with reference to previous studies 14 . In summary, the scoring test comprised an assessment of the accuracy and time required to palpate ve tissues randomly selected for palpation by each participant from 30 tissues around the shoulder joint. To eliminate any selection bias, we used simple randomisation with a forced balance procedure to choose palpation targets a priori. Palpation accuracy was scored as palpation of the correct position of the target tissue without error, with 10 points for a perfect palpation, 5 points for an erroneous palpation at a distance of less than 1 cm, and 0 points for an erroneous palpation at a distance of more than 1 cm. Palpation speed was scored as the time required to palpate the target tissue. Ten points were given if the tissue was palpated in less than 10 seconds, 5 points if it was palpated between 10 and 30 seconds, and 0 points if it was palpated in more than 30 seconds. This scoring was performed for each of the ve tissues, with a maximum of 50 points for accuracy and 50 points for speed, for a total score of 100 points. The following 30 tissues were selected for instruction: Scapula (scapular spine, acromion, scapular spine edge, medial border, superior angle, inferior angle, lateral border, infraglenoid tubercle of the scapula, and coracoid process), clavicle (shaft, acromioclavicular joint, and sternoclavicular joint), humerus (greater tubercle, lesser tubercle, and intertubercular groove), and muscles around the shoulder joint (deltoid, pectoralis major, supraspinatus, infraspinatus, teres minor, teres major, subscapularis, latissimus dorsi, coracobrachialis, trapezius, rhomboid, levator scapulae, pectoralis minor, serratus anterior, and biceps brachii). Students were informed before and after their testing to not share the details of the assessment with other students 19 . The primary outcomes of the study were the accuracy score, speed score, and combined total score.
After the completion of the initial assessment, participants attended a course on palpation of the shoulder joint (90 minutes × 3 sessions in total) at 1, 3, and 5 weeks. To standardise the interventions, the target tissues for instruction in the three lectures were chosen as the 30 tissues as described above. For the US group, the palpation training was conducted by providing visual feedback using US images. Two US imaging systems (ARIETTA Prologue, Hitachi Aloka Medical, Tokyo, Japan and Venue40 MSK, GE Healthcare, Tokyo, Japan) were utilised. The participation schedule was adjusted so that ve participants attended each course 9 . In principle, no tailoring of the course was performed for any participant. Because of the nature of the intervention, we were unable to blind the participants and the administrator. However, to reduce bias, the outcome evaluators were not informed of the participants' assigned group.
One week after the end of the intervention, the students participated again in a scoring assessment of their palpation skills. Further assessment was conducted 12 weeks after the end of the course to con rm the lasting effects (Fig. 1). The scores were also compared between the three assessment time points: (1) pre-intervention (T1) -immediately post-intervention (T2), (2) immediately post-intervention (T2) -12week follow-up (T3), and (3) pre-intervention (T1) -12-week follow-up (T3). During the period from before the start of the intervention to the follow-up assessment, the participants were instructed to not perform any special practical exercises to ensure consistency. To examine the changes within each group and at the time points T1-T2, T2-T3, and T1-T3, comparisons between the corresponding groups were conducted using the Friedman test. If signi cant differences were found, the multiple comparison method (Bonferroni method) was used to examine the changes according to the predetermined time points. At T2 and T3, the Mann-Whitney U test was used to compare the scores between the groups. To examine the amount of change with each intervention and time point, Cohen d effect sizes were calculated for each test that showed a signi cant difference 20 . To manage missing data, simple imputation (mean imputation) was used for both missing data at random and not at random. The sample size was calculated using G*Power 3.1.9.7 (Heinrich Heine University, Düsseldorf, Germany). For the non-paired comparison between groups of non-normally distributed indices, each group was calculated to require 51 participants, assuming an effect size of 0.5, a power of 0.8, and a signi cance level of less than 5%. Due to the impact of coronavirus disease (COVID-19), we were unable to recruit the required number of participants for the sample size. As the pandemic was expected to be prolonged, it became di cult to gather the number of participants initially planned within a short period even if the recruitment period was extended, and thus, the experiment was terminated with the number of participants initially recruited, and the analysis was conducted. In the post-hoc analysis, the power was 0.33. SPSS Statistics 27.0 (IBM Corp., Armonk, NY, USA) was used to perform the statistical analyses.

Results
The recruitment period was 2019-2020, and the follow-up period was up to 3 months following the rst intervention (March 2021). Of the 38 rst-year physical therapy students, 29 (76.3%) participated in the study. It seemed di cult to continue the recruitment of new participants due to the lack of certainty regarding the COVID-19 pandemic. Therefore, this study was suspended in March 2021. The participants were 11 men and 18 women (median age: 20 years old). The baseline scores for each group were similar ( Table 1). The 29 participants were randomly assigned to the two groups, and three participants in the US group and one participant in the control group dropped out during the follow-up period (Fig. 1). Because the study was conducted over 2 years, 2019 and 2020, there was a 1-year time gap between the start of the intervention for approximately half of the study participants. During that time, there were no major changes in other classes or educational curricula. During the intervention, the number of students attending one lecture was limited because of the restriction that up to ve students per teacher were desirable for lectures using US images. Therefore, there was a lag of less than 3 hours between randomisation and the start of the intervention, depending on the group assigned. The actual intervention was performed as planned, with a xed and formulated ow with a short time lag. Finally, 25 participants who were followed up (12 in the US group and 13 in the control group) were included in the analysis. Accuracy score at baseline 20 (15)(16)(17)(18)(19)(20)(21)(22)(23)(24)(25) 27.5  Speed score at baseline 20 (10-25) 22.5 (10)(11)(12)(13)(14)(15)(16)(17)(18)(19)(20)(21)(22)(23)(24)(25) Total score at baseline 40 (30-47.5) 50  Analyses of the assessment scores showed multiple signi cant differences at each time point in the US group (Fig. 2a, b, c). However, no signi cant difference was observed in the control group at any of the time points (Fig. 3a, b, c). In the US group at T1, the median accuracy score was 20 (range, 15-25), the median speed score was 20 (range, 10-25), and the median total score was 40 (range, 30-47.5) (Fig. 2a). In the US group at T2, the median accuracy score was 35 (range, 27.5-40), the median speed score was 25 (range, 17.5-27.5), and the median total score was 37.08 (range, 32.5-45) (Fig. 2b). In the US group at T3, the median accuracy score was 37.08 (range, 32.5-45), the median speed score was 23.75 (range, 20-27.5), and the median total score was 60.8 (range, 55-75) (Fig. 2c). In the control group at T1, the median accuracy score was 27.5 (range, 15-35), the median speed score was 22.5 (range, [10][11][12][13][14][15][16][17][18][19][20][21][22][23][24][25], and the median total score was 50 (range, 20-60) (Fig. 3a). In the control group at T2, the median accuracy score was 27.5 (range, 20-40), the median speed score was 20 (range, [15][16][17][18][19][20][21][22][23][24][25], and the median total score was 47.5 (range, 35-65) (Fig. 3b). In the control group at T3, the median accuracy score was 30.4 (range, 30-40), the median speed score was 25 (range, 20-30), and the median total score was 55 (range, 35-60) (Fig. 3c). The effect sizes for the time points for which signi cant differences in the accuracy score were noted were 0.849 for T1-T2 and 0.990 for T1-T3. The effect sizes for the total score were 0.849 for T1-T2 and 0.849 for T1-T3. Comparison of scores between the groups at T2 and T3 showed no signi cant differences at either time point (Table 2). No adverse events were observed in either group.

Discussion
Our results demonstrated that the US group showed better improvement in skills due to the intervention, while the control group did not show any meaningful improvement. Another nding of this study was that the impact of the intervention in the US group was relevant in the short and medium terms. However, the comparisons of palpation scores between the two groups showed no signi cant differences at each time point.
Many previous studies have supported the use of echo in improving palpation skills. Improved palpation skills have been reported by using echoes at the lumbar spine 11 , shoulder joint 5 , some body landmarks 15 , various upper extremity landmarks 10 , and knee joint landmarks 6 . However, these were only intervention studies without control groups, and the differences between the new method and the usual instructional method could not be assessed.
Several randomised controlled trials have been conducted to overcome the limitations of these studies.
The results of these studies also support the usefulness of ultrasonography, which has been reported to be useful for the shoulder and knee joints 12 , elbow joint, and ankle joints 14 . However, most previous studies examined the immediate effects after the intervention and did not examine the lasting effects. In the present study, to overcome the limitations of previous studies, we conducted a randomised controlled trial with middle-term follow-up from immediately post-intervention to 3 months. To our knowledge, this is the rst study to examine the middle-term effects of the utilisation of US imaging for improving palpation skills. The effect size also indicated a moderate effect.
The results of the scoring data analysis showed a signi cant improvement in the accuracy of palpation, rather than speed, leading to an increase in the overall score. By using US images, students can learn the palpation technique by obtaining visual feedback 5,13 . This might be one of the reasons why the students recruited in our study obtained more accurate palpation skills for each landmark. Palpation using US images is also called 'sonopalpation' 7,8 . Accurate palpation of the anatomical location and accurate examination of the tenderness of soft tissue are very important clinically to identify the tissue causing pain. The importance of palpation skills using US images is expected to increase in the future, and it is desirable to construct a systematic curriculum that includes US imaging education of whole-body landmarks 10 . In contrast, it has been reported that the accuracy of palpation of the foot was still low even after training to improve palpation skills using ultrasonography 19 . In the current study, we mainly targeted the shoulder joint, which has been frequently examined in previous studies on palpation skills using US images. However, it is unclear whether the results of this study can be directly applied to other joints.
There was no signi cant difference between the groups at T2 and T3 (Table 2). Since there was no noticeable difference in the baseline scores, there may not be a relevant difference between these two intervention methods. Considering these advantages, purchasing a new device seems unnecessary. It is also not recommended that a US device being used for a certain purpose, such as abdominal diagnosis and medical check-up, be moved too far to another department that does not have US diagnostic equipment.
The present study has some limitations. First, there was the possibility that the evaluator's judgement was not perfectly accurate. However, a well-trained physiotherapist specialising in sports and musculoskeletal rehabilitation judged the accuracy of the palpation performed by the examinee students. Further, the physical therapists who evaluated the accuracy of palpation in this study regularly delivered lectures on palpation to young clinicians and students at universities and used echoes. Nevertheless, in future studies, it would be better to use ultrasonography to con rm the training accuracy 4 . Second, since only one joint was used in this study, the external validity may have been reduced. A previous study reported low inter-rater reliability when comparing the agreement between 32 palpations of the long head biceps tendon using ultrasonography by two physical therapists with a certain level of experience 4 . In future studies, palpations of different body landmarks should be assessed. Third, participation in this study was voluntary, and thus, there was a possibility of self-selection bias. Further, a perfect participation rate could not be expected. Additionally, the fact that this trial was conducted during a global pandemic may have contributed to the low participation rate. We should have taken all possible safety measures, such as su cient infection control measures, and made announcements so that potential study recruits would have a strong sense of safety. Further studies that address the limitations of the current study are needed to clarify these points.
In conclusion, when compared with usual training courses, US-based training may improve the skills of medical students, especially in terms of the accuracy of palpation. There were no invasive or adverse events associated with the use of US imaging equipment. We thus recommend that mentors use US imaging during palpation training. However, we do not recommend that all educational programmes introduce US imaging devices but that the usefulness of ultrasonography should be evaluated based on the situation. Figure 1 CONSORT ow diagram of the study.  Palpation scores at each time point for the control group.