When assessing outcomes of treatment, researchers typically use questionnaires or scoring systems (also known as assessment tools) to allow for the presentation and publication of data. Ideally, these assessment tools would be standardized and generally accepted for specific populations, so that the outcomes from different studies could be compared. Of equal importance, although rarely done in the past, all assessments should be tested for reliability, validity, and responsiveness before clinical use (3).

Reliability assures that repeated administration of an assessment will give the same results. An assessment has Test-retest reliability if it produces the same results when given different times under the same conditions. Internal reliability occurs when an assessment evaluating similar conditions produces similar results.

Validity assures that the assessments measure outcomes accurately— that is, does the test measure what it was designed to measure? The three approaches used to measure validity include content validity, which determines whether all of the important aspects of a condition are covered by the assessment, criterion validity, which assures that the scores produced by the assessment correlate with accepted standards, and construct validity, which demonstrates that the assessment produces results consistent with the existing understanding of the field or with other assessments.

Responsiveness refers to the ability of an assessment tool to measure changes over time. The responsiveness index is one method of evaluating the responsiveness of an assessment tool and is determined by the mean change score divided by the variability in scores among stable subjects (4,5). Other methods include the standard response mean, which is the mean change in the score divided by the standard deviation of the change scores (6), the relative efficiency, which is the squared ratio of the t statistic (7), and the effect size, which is determined by the mean change score divided by the standard deviation of baseline scores (8). An assessment tool that is highly responsive detects even small changes in outcome over time, increasing the statistical power of the study and allowing researchers to perform studies with fewer subjects.

Of the many available assessment tools developed for evaluating the shoulder, only those developed recently have been assessed for reliability, validity, and responsiveness. Examples include the Self-Administered Questionnaire for Assessment of Symptoms and Function of the Shoulder (9), the Western Ontario Shoulder Instability Index (10), the American Shoulder and Elbow Surgeons Standardized Shoulder Assessment Form (11), the Simple Shoulder Test (12, 13, 14), and the Constant-Murley evaluation (15,16). Clearly, the use of a valid and reliable assessment tool is of great value in the assessment of outcomes for the treatment of the athlete’s shoulder, but the question remains: which of the many assessments should be used?

Assessment tools used in orthopaedic surgery generally tend to fall into a hierarchy of three different levels of sensitivity:

Like the different powered lenses on a microscope, each of these levels of assessment provides a different perspective on the patient’s outcome (17, 18, 19). General health assessments are best suited to discriminate the outcomes of differing treatments for systemic diseases or conditions, such as rheumatoid arthritis. Other conditions (e.g., shoulder pain when throwing) may not affect general health, but rather would be better assessed by disease-specific or population-specific assessment tools. Joint- or extremity-specific assessments are common in orthopedic surgery, and try to provide a method of assessing all potential disorders for a particular joint or extremity. There are a number of global shoulder assessments in the orthopedic literature (Table 5-1). Disease- or population-specific assessment tools are likely the most sensitive because they are designed to apply to a very specific population of patients.

General health assessments report a score for the general health of the patient. The best known example of a general health assessments is the standard form (SF) 36 questionnaire (20), which measures eight different aspects of health, scoring them on a 100-point scale (Appendix 5-1). These aspects of general health include assessment of the following: (a) limitations in physical activities due to health problems, (b) limitations in social activities due to physical or emotional problems, (c) limitations in usual role activities due to physical health problems, (d) limitations in usual role activities due to mental health problems, (e) general mental health and well-being, (f) bodily pain, (g) vitality, and (h) general health perceptions (20). Not unexpectedly, the ability to perform overhead sports is not weighted heavily on this assessment; general health assessments are much more sensitive when assessing the effects of a systemic disease (e.g., arthritis). Other examples of general health assessments include the Arthritis Impact Measurement Scale (21), the Nottingham Health Profile (22), and the Sickness Impact Profile (23)—all of which may be more relevant to orthopedic treatments than the SF-36 assessment (24,25). However, because the SF-36 assessment has been well validated and successfully used to evaluate a number of medical conditions, its use is increasing for the general health evaluation of patients with orthopedic conditions. With regard to the shoulder however, the SF-36 form does not seem to correlate well with other measures of shoulder disability (26,27).

The next level of assessment in measuring outcomes in orthopedic surgery focuses on a particular joint or extremity. Assessments at this level may be more sensitive for evaluating the treatment of orthopedic conditions. A number of different global shoulder assessment methods have been used (Table 5-1). These assessments are designed for the evaluation of most disorders of the shoulder and as such may have some application to overhead athletes. Some global shoulder assessments produce a final numeric grade, which allows statistical comparison of the overall preoperative and postoperative condition of the shoulder and allows for comparison of different methods of treatment (9,15, 16,28, 29, 30). Other assessments serve as guides to assure that none of the important elements in the evaluation of outcome are neglected and, although comprehensive, may be lengthy and cumbersome (29). Yet other assessments for the shoulder take a different approach and are designed to be as simple as possible (12, 13, 14).