While a few studies have investigated the reliability, sensitivity, and specificity of the GALS examination [7, 15, 18, 19], we believe this to be the first study to investigate its use in primary care by comparing the results of the GALS exam between family physicians and rheumatologists. Results of this pilot study revealed a reasonable level of agreement between rheumatologists and family physicians recently taught to perform the GALS examination via an instructional DVD (estimated Kappa = 0.3675; 95% CI: 0.3009, 0.4342, Pobs = 0.698). Upon further analysis of the individual components of the exam, assessments of gait and arm movement were found to have the greatest level of agreement, while the appearance of the legs and spine were identified as the sources of greatest disagreement. Gait is an extremely important component of the GALS exam since its assessment often contributes information with respect to a patients' propensity to falling [20, 21, 21].
To more accurately assess the source of disagreement, positive and negative agreement of all components of GALS were determined. Results revealed that family physicians were more likely to agree with rheumatologists when the trait being assessed was considered normal as opposed to abnormal, as shown in both Tables 3 and 4. Similarly, Hood and colleagues reported greater negative predictive values in the assessment of 200 patients suffering from acute or chronic musculoskeletal conditions, also suggesting that negative or normal traits are more easily identified .
Despite the fact that few studies have examined the use of GALS by different health care professionals, a pattern that has previously emerged, and one that was also noted in the current study, is the variation in the assessment of the appearance of the spine. Plant et al. investigated the reliability of the GALS examination when conducted by senior house officers and registrars in rheumatology (N = 30) and reported the greatest disagreement when scoring of the appearance of the spine . Jones et al. also reported difficulties in the identification of other spinal abnormalities, particularly for lateral cervical flexion . It has been suggested that age-related changes affecting flexibility of the neck and back are the likely source of the difficulties encountered in differentiating normal from abnormal spinal appearance [7, 19]. Thus, it is plausible that these and other age-related changes may also contribute to difficulties distinguishing normal from mildly abnormal traits in other components of the GALS examination.
Further comparisons of our results with those of Plant et al. revealed a similar level of observed agreement; however, the reported reliability (estimated kappa) differed significantly. A well-known and frequently observed trend is that of the relation between reliability and degree of scale complexity (i.e. dichotomous scales, Likert scales etc.) where an increase in the number of possible outcomes (i.e. none, mild, moderate, or severe) results in increased reliability . The decreased level of reliability as assessed by the kappa statistics (min = 0.13, max = 0.49) in this study may, in part, be attributable to the dichotomous nature of the scale employed where appearance and movement could be labeled only as normal or abnormal. In contrast, Plant et al. replaced the traditional dichotomous scale with one that allowed examiners to rate features as normal, mildly, moderately, or severely abnormal and subsequently reported higher kappa statistics varying from 0.49 to 0.74 .
Although the results of the current study appear to suggest that difficulties persist in the recognition of musculoskeletal abnormalities, one should be cautioned about making definitive conclusions without acknowledging factors which may have contributed to or limited the observed level of agreement. For instance, a review of patient scoring sheets completed by family physicians and rheumatologists revealed that while both examiners recognized similar patient characteristics, there was discrepancy between the comments recorded and the identification of these features as normal or abnormal. In a given patient, for example, some physicians recorded gait to be abnormal due to an observed limp, while others also noted the presence of a limp but incorrectly labeled this as normal. This observation helps to explain the trend observed in other features of the GALS exam where rheumatologists consistently labeled more features as abnormal than family practitioners as seen in Table 3. This may be another example of what some physicians may deem normal, age-related changes, thus assessing the feature as normal, while others would assess the feature as being abnormal relative to a healthy standard. This discrepancy may be linked to differences in the perception of abnormalities between family physicians and rheumatologists. However, given the fact that the recorded observations could not be objectively quantified or assessed as being mildly or moderately abnormal as in the study by Plant et al., these differences ultimately resulted in a decreased level of agreement. It is believed that agreement would have improved significantly had the newly trained family physicians been given an opportunity to directly observe the GALS examination as conducted by a rheumatologist and to meet with rheumatologists prior to the study to discuss characteristics that differentiate normal features from those that are abnormal. By coming to a consensus as to how to score certain features (i.e. the limp), it is anticipated that agreement would have been higher. In addition, variation in scoring between family physicians and between rheumatologists was not assessed. Characteristics of the cohort can also influence the measures of agreement, particularly the kappa statistic. For instance, the lack of symmetry in the study population (i.e. the majority have a musculoskeletal condition) will tend to produce lower kappa values [23, 24]. One of the major limitations to this study was the asymmetry in the study population which consisted of only 7 participants who had never been identified with any musculoskeletal conditions by the ICD codes.
The prevalence of MSK abnormalities in this ambulatory study population was also estimated for each anatomical region. These were further subdivided by the joints that were involved. It was apparent that the most common features assessed as being abnormal by the rheumatologists were those in the joints of the fingers/hands (20.8% of patients) and the toes/feet (19.2% of patients). The vast majority of these abnormalities were cases of osteoarthritis in the peripheral joints, none of which had previously been documented in the patients' family practice charts. Decreased cervical range of motion (9.6% of patients) and abnormal knees (8.8%) were also prevalent in this population and were regions that had not been documented as abnormal by the patients' family physicians. There may be a couple of reasons for these "new" abnormalities; a) lack of documentation by the family physician, b) the patient has experienced these problems but not expressed/reported them to his/her family physician. The majority of these newly identified abnormalities would require further investigation (i.e. kyphotic posture being assessed for osteoporosis) or treatment (swollen joint treated with medication). Only one other study has used the GALS exam to investigate the prevalence of MSK abnormalities. This study was conducted in acute and chronic medical in-patients . Here the GALS screening tool was positive (abnormality identified) in 53% of acute patients and 94% of chronic patients where osteoarthritis accounted for the majority of rheumatological conditions identified in the both study populations.
A future study will include a wider variation in subject ages so as to obtain a sample population without any previous musculoskeletal diagnoses allowing the sensitivity and specificity of the GALS exam to be investigated. This study will also involve the analyses of subgroup of patients who are assessed by all family physicians and all rheumatologists to assess the inter-observer variation. In addition, these results also suggest that an instructional DVD alone may not be the most effective and consistent method of teaching the GALS exam but that the DVD should be accompanied by oral instruction/interaction. This may be more important when instructing physicians who have already developed their skill set or routine as compared to medical students who have little to no background in this area.
Although the ability of family physicians to assess the MSK system prior to the introduction of the GALS examination was not assessed, our results suggest that family physicians can efficiently use the GALS examination to assess the MSK system, by integrating it into their routine physical exam. Nevertheless, previous studies of medical professionals whose ability to assess the MSK system before versus after learning the GALS examination was evaluated have revealed that physicians' confidence and efficiency in examining the system had increased significantly [13, 14]. These results suggest that the same may be true for the family physicians of the current study.