Subjective and Objective Numerical Outcome Measure
Assessment (SONOMA). A Combined Outcome Measure Tool:
Findings on a Study of Reliability

This section was compiled by Frank M. Painter, D.C.
Send all comments or additions to:   Frankp@chiro.org

FROM:   J Manipulative Physiol Ther 2003 (Oct); 26 (8): 481–492 ~ FULL TEXT

B. Timothy Harcourt, DC, Manel Wijesinha, PhD, Gary E. Harcourt, DC

Pennsylvania State University, York, USA

---

OBJECTIVE:   To determine the reliability of a combined tool, namely that of Subjective and Objective Numerical Outcome Measure Assessment (SONOMA).

METHODS:   Testing was conducted, limited to patients with neck, midback, or lower back pain, with or without radiculopathy, in an outpatient chiropractic office setting. Test-retest reliability of the objective analysis of SONOMA was carried out on the same day (n = 50) with an interval time period of less than 60 minutes. Between-day reliability of the subjective analysis of SONOMA was carried out with an interval time period of 24 hours (n = 50). Individual and combined parameter reliability was established for the tool.

RESULTS:   Short-term objective and between-day subjective reliability coefficients were high. The Pearson correlation coefficient for the combined tool was .96, and the coefficient for the individual parameters ranged from .55 through .93. All these correlations were statistically significant, with a P value not more than .0001.

CONCLUSIONS:   The SONOMA combined numerical outcome measure tool demonstrated a high degree of reliability. This outcome tool measures directly and therefore reflects patient pain perception, functional status, and provider-driven objective assessment. We believe this tool provides the unique combination of both subjective and objective functional capacity assessment. It should be valuable for day-to-day practical application, as well as considered for future clinical trials and quality-of-care studies. This combined tool shows promise as having a high degree of reliability and, hence, may demonstrate a comprehensive representation of the patient-clinical picture, particularly in regard to functional capacity assessment.

---

From the FULL TEXT Article:

Discussion

      Simplicity and practicality

Modern methods of outcome assessment or functional capacity assessment must be simple, practical, economical, reliable, and valid. [20] Patients were able to complete the visual analog scale with superimposed NRS along with the activities of daily living questionnaire, otherwise known as the subjective analysis, in 3 to 5 minutes. The objective analysis, consisting of muscle strength, range of motion, joint dysfunction, additional findings, and pressure pain thresholds, was typically obtained per examiner in 30 minutes or less. The straightforward nature of numerical calculation, both within the subjective and objective analyses, demonstrated its designed simplicity. We believe that because this tool involves separate measurements of pain perception, activities of daily living, and physical parameters, it provides a more diversified and clearer representation of the patient clinical picture.

      Reliability

For purposes of statistical analysis, Nilsson et al [26] noted that correlation coefficient values between .80 and 1.0 indicate a high degree of reliability, and a correlation coefficient between .60 and .79 indicates moderate reliability. Additionally, Troyanovich et al [27] highlights that intraclass (in this study intratool or interexaminer) correlation coefficients of greater than .70 are considered accurate enough for use in clinical and research applications.

      Subjective analysis

The intratool Pearson correlation coefficient for the VAS with superimposed NRS was r = .92 (P < .0001). This demonstrates high reliability, despite a 24-hour test-retest time interval. The test-retest 24-hour time interval was chosen in an effort to minimize short-term memory bias and to maximize sensitivity and accuracy testing of the VAS with superimposed NRS. Additionally, it should be noted that a very low P value indicates the reliability for this parameter to be highly statistically significant.

The intratool Pearson correlation coefficient for the ADL questionnaire was r = .93 (P < .0001). Again, this demonstrates high reliability in light of a test-retest 24-hour time interval in an effort to minimize short-term memory bias and to maximize sensitivity and accuracy testing of the ADL questionnaire. Additionally, it should be noted that a very low P value indicates the reliability for this parameter to be highly statistically significant.

The intratool Pearson correlation coefficient for the subjective analysis, which includes the visual analog scale with a superimposed NRS and the activities of daily living questionnaire was r = .92 (P < .0001). This, again, demonstrates high reliability for the entire subjective analysis package in light of a test-retest 24-hour time interval in an effort to minimize short-term memory bias and to maximize sensitivity and accuracy testing of the ADL questionnaire. Moreover, it should be noted that a very low P value indicates the reliability for this parameter to be highly statistically significant.

      Objective analysis

The interexaminer Pearson correlation coefficient for the physical parameter of muscle strength was r = .80 (P < .0001). Additionally, it should be noted that a very low P value indicates the reliability for this parameter to be highly statistically significant. This represents high reliability and stands in stark contrast to previous correlation studies for physical parameters as noted by Deyo. [3] We believe it is desirable to improve reliability of a tool by simplification, as long as it is not to the detriment of responsiveness. We believe one of the factors involved in this high reliability value is simplification of muscle strength evaluation, being that of either normal strength or abnormal strength (dichotomous scale approach to physical parameters). Moreover, we believe responsiveness is not lost, as the determination of normal or abnormal strength was obtained through an ordinal scale of muscle strength described by Hoppenfeld. [21]

The interexaminer Pearson correlation coefficient for the physical parameter of range of motion was r = .86 (P < .0001). This demonstrates high reliability and stands in stark contrast to previous correlation values for physical parameters as noted by Deyo. [3] Also, it should be noted that a very low P value indicates the reliability for this parameter to be highly statistically significant. Once again, we believe the improvement in reliability values may be due to the designed dichotomous approach to physical parameters, with range of motion being evaluated as either normal or abnormal. Additionally, range of motion measurement was standardized with both the initial and follow-up examiners using the same tools for analysis.

The interexaminer Kendall t correlation coefficient for the physical parameter of joint dysfunction was ? = .68 (P < .0001). This represents moderate reliability but still represents a higher reliability value than previously analyzed physical parameters, particularly with respect to joint dysfunction. Additionally, it should be noted that a very low P value indicates the reliability for this parameter to be highly statistically significant. We believe that the significant reduction in weightedness with regard to joint dysfunction, by allowing the examiner to palpate for either the presence or absence of joint dysfunction within a certain spinal region, significantly improves the reliability value. This reliability value is less than all parameters noted above and would be consistent with what would be expected for the physical parameter of joint dysfunction versus other parameters. [28-29] In other words, in light of previous reliability studies, we would expect the relative reliability of the joint dysfunction parameter to be lower.

The interexaminer Kendall t correlation coefficient for the physical parameter of joint dysfunction was K = .68 (P < .0001). This demonstrates moderate reliability, but at the same time, a very low P value indicates the reliability for this parameter to be highly statistically significant. The relatively subjective nature of evaluating the presence or absence of antalgia, antalgic gait, or nerve root compression most likely contributed to a lower correlation coefficient value.

The interexaminer Pearson correlation coefficient for the physical parameter of pressure pain thresholds was r = .55 (P < .0001). It should be noted that the test-retest time period was less than 60 minutes (generally 10 to 15 minutes), and possible “irritation” by the initial examiner could potentially lower thresholds for the second examiner. Despite the short test-retest time period, moderate reliability was demonstrated for this physical parameter. Additionally, it should be noted that a very low P value indicates the reliability for this parameter to be highly statistically significant.

The interexaminer Pearson correlation coefficient for the objective total, which encompasses muscle strength, range of motion, joint dysfunction, additional findings, and pressure pain thresholds, was r = .87 (P < .0001). This demonstrates high reliability and is highly statistically significant and consequently speaks to the importance of simplicity and diversification when it comes to quantifying physical parameters.

The Pearson correlation coefficient for the combined total of SONOMA was r = .96 (P < .0001). This demonstrates high reliability, as well as being highly statistically significant, for the combined tool, which encompasses both the subjective and objective analyses. Once again, at first glance, this reliability value seems higher than what would be expected given individual correlation values. However, a deeper look reveals that even though initial and follow-up examiners may obtain slightly different values in the different subcategories, the end result carries an extremely high reliability correlation value or reliability. We believe this strongly speaks to the issue of simplicity, yet diversity and reliability. We also believe that pain perception, function, and physical parameters should be measured directly. When these categories are measured directly and combined into a numerical total, reliability can be high, as demonstrated by the SONOMA tool. Lastly, we believe that because this tool employs separate measurements of pain perception, activities of daily living or function, and physical parameters, it provides a diversified and clear representation of the patient clinical picture. This simple yet diverse quantitative clinical picture is more reliable than individual outcome measures standing alone.

---

Conclusion

To our knowledge, the SONOMA tool represents the first outcome measure tool that evaluates pain perception, activities of daily living or function, and physical parameters separately and combines values for a reliable and diversified depiction of the clinical picture. A very high correlation coefficient of .96 (P < 0.0001) demonstrates the reliability of this simple and practical tool. It is simple and practical for both the patient and the doctor.

Return to OUTCOME ASSESSMENT

Since 5-14-2004