General Comments

First, I would like to congratulate Magnan and Maklebust on a job well done. This study was carefully constructed and reported and the findings were as interesting as they were unexpected. It reminds me of a quote by Isaac Asimov, a scholar and a science fiction novelist, who said, "The most exciting phrase to hear in science, the one that heralds new discoveries, is not 'Eureka!!' but 'That's funny [horizontal ellipsis][horizontal ellipsis].'" At this point, I can only comment on which of these results are "funny" and what some of the potential reasons might be.

Placed in the Context of Previous Reports

First, looking to place these findings in the context of previous reported interrater reliability studies, what might account for differences? I can speak most knowledgeably about the initial group of studies1 reported in 1987, which most closely resembles the Magnan and Maklebust study.2 In these cases, I believe differences may have occurred because our studies were done in long-term care settings where the caregivers and the "expert" rater had cared for the patients being rated for many weeks. This degree of familiarity sharpens the precision of the rating for all involved and consequently sharpens interrater reliability. The effect of increasing familiarity with the patient being rated can be seen in improved predictive validity as well, so that ratings done 48 hours following admission in tertiary care are found to be more strongly predictive of pressure ulcer development than ratings performed at the time of admission.3

Unfortunately, given the rapid turnover of patients in the acute care setting, this degree of familiarity with any patient is now quite unusual. This lack of familiarity leads to a bit of guessing, which may account for some loss of precision and interrater reliability. This is probably one of the factors at play in this study, where some participating nurses were pulled from neighboring units to rate a patient.

Considering Differences Between New and Regular Users

Magnan and Maklebust's explanation that the regular users tend to become "habituated" to using clinical judgment rather than a strict constructionist use of the Braden Scale is probably apt. In one study comparing clinical judgment of patients' functional status to formal assessment of functional status, investigators found that physicians and nurses using clinical judgment alone correctly identified severe impairment. In contrast, moderate impairments in mental status, nutrition, vision, and continence easily identified with formal assessment were poorly recognized using clinical judgment alone.4 Likewise, the new users in Magnan and Maklebust's study, without the intrusion of habituated assessment practices, were more reliable and precise in identifying patients at mid-levels of risk after being trained in the use the Braden Scale.

Another problem besides the "habituation" of clinical judgment may be at work here. I believe that regular users eventually begin to "drift" into interpretations of the subscales that are at variance with the written descriptors and these erroneous interpretations become habituated. These interpretations may not be obliterated by some types of retraining. New users who have formed no rating habits (let alone bad habits) are much more amenable to the effects of training.

Lastly, the Web-based training module uses exact wording from the Braden Scale in the case studies. This helps the staff achieve good reliability during the online exercise, but it may not help them when they are faced with the ambiguities that come with complex clinical situations seen in their practice.

Practical Considerations

Most learning occurs through a process of input -> practice -> feedback. Input involves the learner receiving information from a source while practice involves having the opportunity to apply that information to a situation. Feedback occurs when the learner receives information on the accuracy of his or her performance during the practice opportunity.

Having used the Web-based training tool, I can attest that it provides excellent opportunities for practice with case studies and feedback through the self-correcting mechanism. Upfront input on the Braden Scale is minimal in the Web-based training program, however, and most of it is embedded in the case studies. It is possible that both regular users and new users might benefit from more in-depth input on the meaning of each subscale of the Braden Scale, before entering into the practice and feedback phase. This might be particularly true for regular users whose interpretation may have begun to drift away from the intended meaning.

This input could be provided in a number of formats. Either of two articles I wrote soon after developing the Braden Scale could be used.5,6 There are videotapes on the Braden Scale available commercially, including one sold on the Web site (http://www.bradenscale.com). Used alone, any of these resources would provide only input, and therefore will be incomplete. If followed by the Web-based training, the input -> practice -> feedback loop would be completed.

References