Presentation is loading. Please wait.

Presentation is loading. Please wait.

Is the Berg Balance Scale an Internally Valid and Reliable Measure of Balance Across Different Etiologies in Neurorehabilitation? A Revisited Rasch Analysis.

Similar presentations


Presentation on theme: "Is the Berg Balance Scale an Internally Valid and Reliable Measure of Balance Across Different Etiologies in Neurorehabilitation? A Revisited Rasch Analysis."— Presentation transcript:

1 Is the Berg Balance Scale an Internally Valid and Reliable Measure of Balance Across Different Etiologies in Neurorehabilitation? A Revisited Rasch Analysis Study  Fabio La Porta, MD, Serena Caselli, PT, Sonia Susassi, PT, Paola Cavallini, PT, Alan Tennant, PhD, Marco Franceschini, MD  Archives of Physical Medicine and Rehabilitation  Volume 93, Issue 7, Pages (July 2012) DOI: /j.apmr Copyright © 2012 American Congress of Rehabilitation Medicine Terms and Conditions

2 Fig 1 Targeting (person-item threshold distribution) graph of the BBS-12. Observations (N=302) and item thresholds are displayed, respectively, in the upper and the lower part of the graph, separated by the logit scale. Grouping set to interval length of .20 making 65 groups. Archives of Physical Medicine and Rehabilitation  , DOI: ( /j.apmr ) Copyright © 2012 American Congress of Rehabilitation Medicine Terms and Conditions

3 Fig 2 The BBS-12 ruler. In order to use the ruler, the BBS is administered according to the original BBS scoring criteria, and the original score for each and every BBS item is reported in the corresponding blank square on the left hand side in the [A] area. The 2 items excluded from BBS-12 (BBS02–standing unsupported and BBS03–sitting unsupported) are not administered. The BBS items are ordered by progressively increasing difficulty from top to bottom. The numbered horizontal bars for each item [B] represent the range of ability flagged by each score, and the actual affirmed score for each item is indicated by a black bar. Thus, for instance, an affirmed score of 4 in BBS05 (transfers) flags an ability range from around +0.5 logits to infinite, whereas an affirmed score of 3 in BBS04 (from standing to sitting) covers an ability range from around −2 to +0.5 logits. For the rescored items, the rescoring key (showed also in table 2) is presented as a number in square brackets. Thus, for instance, no rescoring key is provided for the BBS05 item, and hence, a score of 4 remains unchanged for this item. On the other hand, the rescoring key for the BBS06 item suggests that either a BBS original score of 1, 2, or 3 flags the same ability range indicated by a BBS-12 score of [1]. In the [C] area, either the modified BBS-12 scores or the unmodified BBS scores are reported as required, and their sum is then reported in the blank square in the [D] area. Thus, for this individual patient, the BBS-12 total score is 21. This score is marked both on the top and bottom score rulers [E]. The 2 markers are then conjoined with a vertical dotted line [F] that is the measurement line. This line crosses the measure ruler [G], giving the corresponding Rasch measure estimate expressed in logits (in this case, +0.5 logits). In the area below the measurement ruler, there is a graph [H] plotting the confidence interval due to measurement error around each BBS-12 total score. Then, 2 lines parallel to the measurement line are plotted considering the lower and upper 95% confidence interval around the person's ability [I], thus defining a measurement area [L] that contains the true measure estimate with a 95% confidence. In this way, the true ability estimate for this person lies in the range from −0.4 to +1.4 logits (or 0.5±0.9 logits if expressed as 95% confidence interval). By examining the relationship between the measurement line and measurement area with the individual responses to each item, it is possible to observe several patterns, giving useful clinical information. In some cases (a), the actual response to the item spans across the whole measurement area, whereas in other cases the item response crosses, either from the right-hand side (b) or from the left-hand side (c), the measurement line, although it overlaps just partially the measurement area. These patterns can be considered “best” responses, as the range of ability flagged by the item score is the closest possible to the individual's range of ability flagged by the measure. However, for other items, less expected responses could be observed. For instance, for some items (d), the score overlaps the measurement area just partially without crossing the measurement line. For this item (BBS13–tandem standing), the best response would have been a modified BBS13 score of [1] (that would cross the measurement line), thus indicating that in this activity, this patient underperformed somehow given his ability level. A similar consideration may be applied to the responses to some items that are completely outside the measurement area (e): in this specific item (BBS14) and in BBS12, the patient performed better than expected on the basis of his ability level. The (d) and (e) responses, although unexpected, are still compatible with the model's expectations because the range of ability flagged by the actual item score is still quite close to the measurement area. However, the same consideration may not apply for the response to the BBS01 item (sitting to standing), for which the lowest score is reported (f). Indeed, this is a rather unexpected response (as for this item the best response would have a modified BBS01 score of [3]) that is rather distant from the measurement area. This represents a significant departure from the model's expectations: although this patient's balance ability is in the moderate to high range, nonetheless he fails a rather easy item. Given this unexpected response, a careful inspection of this patient's record is mandatory in order to find an explanation for this unexpected item behavior (transcription error? correct response due to specific individual factors?). Archives of Physical Medicine and Rehabilitation  , DOI: ( /j.apmr ) Copyright © 2012 American Congress of Rehabilitation Medicine Terms and Conditions


Download ppt "Is the Berg Balance Scale an Internally Valid and Reliable Measure of Balance Across Different Etiologies in Neurorehabilitation? A Revisited Rasch Analysis."

Similar presentations


Ads by Google