[Editor's note: Both the letter to the editor by Franchignoni and Giordano and the response by Padgett and colleagues are commenting on the accepted but unedited author manuscript version of this article that was published ahead of print on June 7, 2012.]
We thank Franchignoni and Giordano for their insightful comments1 on our article.2 We would like to initially acknowledge that we agree our study represents preliminary findings. We did not intend for our preliminary study to answer our title question: “Is the BESTest at its best?” Rather, we sought to propose an alternative assessment that would allow the question to be formally tested by future larger-scale comparative studies. We also agree that the 2 analyses used to derive the Mini-BESTest and Brief-BESTest are complementary, with neither producing definitive evaluations alone. This point of view represents the very reason we hoped our analysis would add value to the goal of developing a feasible, reliable, and valid clinical balance assessment.
Following our study, at least 3 fundamental questions remain:
-
Which test version produces the most useful clinical tool for insights regarding sensitivity to change, as well as risk of falls, comorbidities, or decrement in community participation and quality of life?
-
How generalizable are such insights across populations with different clinical diagnoses?
-
Is there added value to including items that represent all of the BESTest's theoretical contexts of balance performance?
These questions will require summative evidence across multiple studies from independent research groups, but the letter by Franchignoni and Giordano, as well as some of our own data, provide some initial insights to these questions. We now will respond to 5 concerns highlighted by the letter.
First, Franchignoni and Giordano note that one item may not be sufficient to assess each theoretical system for the purpose of directing interventions toward a specifically impaired dimension. We agree that subscoring the Brief-BESTest would not be appropriate. Instead, it may be more useful to consider the examination as a balance assessment based on a systems approach that seeks to evaluate balance impairment across multiple contexts of task performance. With this theoretical perspective, there is no assumption that distinct underlying systems of postural control exist, thereby rendering it unnecessary to statistically prove their existence as separate factors. The goal instead becomes to comprehensively assess multiple contexts of balance performance. Therefore, statistically separating items into different factors would support developing an examination that ensures inclusion of each factor (perhaps decreasing the number of items within each factor for brevity as opposed to removing factors) in order to ensure all contexts of task performance are assessed. A systems approach then would represent the basis of treatment, rather than an impairment-based approach as our study might have unintentionally implied.
Second, Franchignoni and Giordano note that the Brief-BESTest was not replicated in their own analysis. It would be interesting to know which items were derived from their analysis. Furthermore, given the complexity of the Mini-BESTest's derivation,3 it then must be asked: Would 2 independent data sets scored from independent research groups also replicate the Mini-BESTest? Is replicating either version of the examination even a useful endeavor, or would a comparative evaluation of the proposed versions better serve the clinical and scientific community?
Third, Franchignoni and Giordano provide an interesting analysis to quantify item difficulty in order to ensure sensitivity across varied levels of performance when developing a new evaluation. Their analysis indeed suggests that future comparative studies will need to evaluate the risk for ceiling and floor effects between the Mini-BESTest and Brief-BESTest. Our preliminary findings, however, do not suggest any increased risk of such effects based on the score distributions, and we found that the easiest or hardest items added very little to the examination due to a lack of variance in the data. Also, as noted in our article's discussion,2 the higher levels of accuracy produced by the Brief-BESTest appear to be due, in part, to the removal of insensitive items. We concede that this lack of variance may be a function of the study sample; however, interestingly, Franchignoni and Giordano's item-difficulty analysis appears to confirm our findings that the Brief-BESTest includes items of moderate difficulty that exhibit differences in performance across participants. Nevertheless, further study on a more diverse study sample would be necessary to determine whether ceiling or floor effects hinder the diagnostic or prognostic value of the Brief-BESTest.
Fourth, Franchignoni and Giordano also identify local dependence as a potential concern because 2 of the tasks in the Brief-BESTest are evaluated twice (once each on the left and right sides). We ran a logistic regression on the second cohort in our study2 to identify individuals with and without a recent fall history, but in this instance we evaluated an examination version of only 6 items in the Brief-BESTest, such that lateral push-and-release scores and one-leg stance scores were included only from the most impaired side. One “nonfaller” became misidentified as a “faller” in this new analysis. This very cursory analysis suggests the Brief-BESTest might retain a high level of diagnostic accuracy for the fall status of individuals with and without multiple sclerosis, but accuracy might be weakened when not accounting for potential lateral asymmetries in balance impairment. Although risking local dependence on evaluating the postural adjustment and postural response, retaining bilateral evaluations may prove most useful when assessing individuals with diseases that progress from unilateral to bilateral impairment (eg, Parkinson disease).
Lastly, Franchignoni and Giordano highlight our third question posed above regarding whether items that represent mechanical constraints and limits of stability improve the diagnostic value of the Brief-BESTest. To further explore this question, we created 2 control versions of the Brief-BESTest: a 6-item version without the items representing mechanical constraints and limits of stability (ie, removing items on hip abduction and forward reaching) and an 8-item version that replaces the items representing these 2 contexts of performance with items from already-represented contexts (ie, replacing hip abduction and forward reaching with rise-to-toes and backward compensatory stepping, the items on anticipatory postural adjustments and postural responses with the next highest item-total correlations). Compared with the Brief-BESTest's 100% accuracy, the 6-item control version elicited 71% sensitivity, 95% specificity, and an accuracy of 88% to differentiate participants based on their reported fall history; the 8-item control version elicited 86% sensitivity, 95% specificity, and an accuracy of 92%. The results, therefore, preliminarily suggest that including all contexts of balance performance may be important to the utility of a clinical balance assessment such as the Brief-BESTest.
In summary, we thank Franchignoni and Giordano for their insightful comments and analysis. Their letter highlights the complexity of designing a clinical assessment and raises further questions regarding the Brief-BESTest's and Mini-BESTest's psychometric properties. Although the statistical evaluations of factor loadings and item difficulty are of interest, our theoretical perspective and our preliminary findings would suggest a need to retain items that represent all contexts of task performance, and that items of moderate difficulty provide highly representative values of subsection scores for each context. We hope our initial study to develop the Brief-BESTest provides an additional tool for future comparative studies with other clinical balance assessments in order to advance our shared goal to provide the greatest diagnostic and prognostic value to the clinician so that we may better serve patients.
Footnotes
-
This letter was posted as a Rapid Response on July 12, 2012, at ptjournal.apta.org.
- © 2012 American Physical Therapy Association