Hypothesis
1.  Somatosensory calibration training improves the ability to determine evenness on standardized paired blocks.
2.  Somatosensory calibration training will increase intra-examiner reliability in determining evenness on standardized paired blocks.
Statistical significance will be measured by using p ≤ .05.

Abstract
Objective / Purpose:
Osteopathic medicine practitioners require reliable manual palpatory skills for identifying structural abnormalities (asymmetries).  Past osteopathic and non-osteopathic palpation training studies have shown poor inter-rater reliability and intra- rater reliability.   The objective of this study is to determine if a fabricated training tool (symmetrical) can help calibrate the somatosensory system.  The study looks at whether this tool improves a clinician’s ability to detect smaller differences in asymmetry and increase inter-reliability and intra-reliability.
Methods:
Testing and training will be performed by participant students from the year one through five classes at the Canadian College of Osteopathy (CCO).
The study will include two groups (experimental condition group, and control group) and two sampling times (test 1 and test 2).  Each sampling will consist of a series of eighteen symmetrical and asymmetrical paired blocks (differences of 0-1.25mm at increments of .25mm) presented in a pre-determined sequence.  A baseline score will be out of eighteen on test 1.   In theory, training will improve the condition group’s accuracy and reliability in finding more symmetrical blocks than the control’s ability.
The two groups consist of an experimental treatment group and a control group.  The conditioned group will be given a symmetrical block training to calibrate on over a period of a month.  After training, the participants’ abilities to differentiate between paired blocks will be re-evaluated and compared.

Results:
This study initially included 68 students, but 55 students completed the study. Each student sampled each block twice (18 blocks presented in the same order). The conditioned group included 24 students and the control group consisted of 31 students. The students were assigned randomly to study group by a coin toss.
Data analysis was done using (SPSS version 10 and Excel 2007).  A comparison between the means showed the pre-condition improved more on test 2 than the control group (independent samples t test p=.037).  The conditioned group improved by a mean of .42 block (from test 1 to test 2) while the control group participant’s performance declined by a mean of 1 block. The control group participants did do worse on test 2, and this was not a chance event. The conditioned group members were 2.5 times to improve on test 2 than were the control group participants. The condition group’s performance was negatively correlated with asymmetry (the less symmetrical, the greater the improvement) from test 1 to test 2 (p=.008). There was no such relationship in the control group. The condition group did worse in reliability in that they changed their answers in identifying more symmetrical blocks in test 2. The training improved performance on the second test specifically for the blocks closer to symmetrical. Reliability analysis showed Cohen’s kappa ranged from .29 to .36 just above chance, which is consistent with the literature. While the analyses in this study all point to training having an effect, this effect was not particularly large. It is not clear whether training on plastic blocks can improve clinical ability. The consistent analyses results and reliability findings that are consistent with the literature findings support internal validity.
Key Indexing Terms
Palpation, model(s), osteopathy, symmetry, asymmetry, reliability, intra-examiner reliability, inter-examiner reliability, Osteopathic Manipulative treatment, Chiropractic.


Conclusion
The conditioned group improved more on test 2 than did the control group (independent samples t-test, p=.037).
Paired t-test results confirmed that the two groups “improved” differently from test 1 to test 2. While the control group participants did worse on test 2 (paired t-test, p=.02), the conditioned group did not do worse. In fact, the conditioned group did better on test 2.
Though the conditioned group showed positive mean improvement of .42 blocks from test 1 to test 2, the control group participants’ performance worsened by an average of 1 block.
Since the control group did worse on test 2 and this decline was unlikely to be a chance event; it is possible that test 2 might have been more difficult for both groups for unknown reasons and that the conditioned group would have also been expected to do worse on test 2. If timing and stress were a factor that diminished both group’s performance, it is possible that the conditioned group might actually have performed much better than they did, and possibly showed a significant difference.
When improvement was categorized as a yes-no variable; it was shown that subjects in the conditioned group were 2.5 times as likely to improve on the second test than were control group participants (58% versus 23%).
When looking at analyses by block asymmetry; the conditioned group’s improvement on test 2 was negatively correlated with asymmetry. The less the block asymmetry, the greater was their improvement from test 1 to test 2 (r = -.53, p=.008). The control group showed no such relationship (r = -.13, p = .48). This pattern is consistent with training improving performance for blocks of low symmetry in the conditioned group
The conditioned group tended to do worse than the control group participants in terms of reliability (same results for test 1 and test 2 for the same block) for blocks of low asymmetry but their reliability was more similar to that of the control group participants for blocks of greater asymmetry.
The data is consistent and supportive of training improving performance on the second test, particularly for symmetrical blocks. All of the statistical evidence taken together offers support for this position.
There was only marginal confounding in this study of the relationship of training and the performance difference between test 1 and test 2.
The reliability analyses showed matched results on test 1 and test 2 that were well above chance but that were not particularly impressive.. Cohen’s kappa that ranged from .29 to .36 indicates fair reliability at best. These results are consistent, however, with typical literature findings ( (Holmgren & Waling, 2006; Hart, 2006; Kmita & Lucas, 2008; Seffinger, Najm, Mishra, Adams, Dickerson, & Reinsch, 2004; Stovall & Kumar, 2010)).
The external validity (generalizability) of this study can be applied to other studies that use models. How this extends to practicing osteopaths is difficult to assess, since the findings from this student group, and may not be replicated in a clinical setting.