January 10th, 2018
Predicting Pressure Ulcer Risk – Validity and Reliability of the Braden Scale
According to the American National Pressure Ulcer Advisory Panel, an ideal tool to assess at-risk population is that which is easy to use, has a good predictive value, has a high sensitivity and has a high specificity (Smith et al., 1995 as cited in Defloor & Grypdonck, 2005, p. 374). Since all at-risk assessment skills are not difficult to use and prior training on use is provided before application, validity of any assessment tool is best assessed via evaluation of its sensitivity and specificity (Defloor & Grypdonck, 2005). Whereas sensitivity refers to the extent to which a diagnostic tool can correctly identify the presence of a condition under consideration, specificity refers to the extent to which the tool correctly fails identify cases without the condition i.e. the test negatives in the population that does not have the condition (Defloor & Grypdonck, 2005). Predictive value indicates the proportion of times that the diagnostic tool returned a true value i.e. proportion of positives when condition is present and proportion of negatives when condition is absent (Predictive value theory, n.d.). This section thus considers the literature that has evaluated the reliability and validity of Braden Scale on such perspectives.
VandenBosch, Montoye, Satwicz, Durkee-Leonard and Boylan-Lewis carried out a study evaluating Braden scale’s predictive validity in 1996. Using a sample of 103 randomly selected patients from general care, intensive care and rehabilitation unit of a 550-bed tertiary care training hospital, and a cut-off point of 17, the study found the Braden scale to have a sensitivity of 59% and specificity of 59% (p. 80). Following such findings, VandenBosch, et al. (1996) recommended the use of “frequent and thorough skin assessment practices” together with the Braden scale since the scale may not predict all cases of individuals at-risk of developing pressure ulcers. In a different study assessing such validity in a nursing home, and using a cut-off score of 18, the sensitivity of Braden scale was noted to be 79%, specificity 75% , a positive predictive value of 54% , a negative test predictive value of 90% and a correct classification rate of 75% (Braden & Bergstrom, 1994, p. 459). Comparable results have been found in a more cent study based in brazil where two cut-off points (18 and 17) were used. In this study by de Souza, Santos, Iri and oguri (2010) sensitivity and specificity under the cut-off score of 18 were 75.9 % and 70.3% while those under a cut-off of 17 were 74.1% and 75.4% respectively (p. 95).
Comparing Norton and Braden scales, Defloor and Grypdonck (2005) do not find favorable results for use of both scales. Using a study sample of 1772 with 314 randomly assigned to a “turning” group, and 1458 to a “non-turning” group, and a cut off point of 12 for Norton scale and 17 for Braden scale, the study found out that as much as 80 % of the patient would get false positives, resulting in unnecessary application of preventative measures (p. 373). Another comparative study of Braden scale, Cubbin and Jackson scale and Douglas scale, using a sample drawn from patients in an ICU unit of a teaching hospital (Seongsook, Ihnsook & Younghee, 2004), presented various findings with respect to sensitivity, specificity, and positive and negative predictive values. Of the three scales Braden Scale (with a 97% and 95% respective rating) was unfavorably rated compared to Douglas scale (100% and 100% respectively) in respect of sensitivity and negative predictive value but favorable in comparison to Cubbin and Jackson scale (89% and 92% respectively) (Seongsook, Ihnsook & Younghee, 2004, p. 199). In terms of specificity and positive predictive value, Braden scale rated favorably (26 % and 37% respectively) to Douglas scale (18% and 34% respectively), but unfavorably to Cubbin and Jackson scale (61% and 51% respectively) (Seongsook, Ihnsook & Younghee, 2004, p. 199). Such findings buttress the need to incorporate other assessments for risk of pressure ulcers as advised by VandenBosch et al. (1996). Despite some of the unfavorable findings on validity of Braden scale in provision of care to older adults, its application continues to be recommended by various organizations providing care to the elderly (e.g. Ayello, 2007; Frantz, 2004; AHRQ as cited in Defloor & Grypdonck, 2005, p. 374).
In terms of reliability, a study by Kottner and Dassen (2008) evaluated the interrater reliability of Braden scale with sample drawn form two nursing homes. From the study, differences in the overall rating by nurses ranged from 0-9 points (Kottner & Dassen, 2008, p. 1501). Expressed according to intraclass correlation coefficient, interrater reliability of the scale ranged from 0.73 to 0.95 at 95% confidence interval (Kottner & Dassen, 2008, p. 1501). For individual items of the scale, such interclass correction coefficients at 95% confidence interval ranged from 0.06 to 0.97, the lowest values being observed in “sensory perception” and “nutrition” (Kottner & Dassen, 2008, p. 1501). Such reliability was not moderated by the work experience of the nurse (Kottner & Dassen, 2008). Reliability in other studies have been reported to range between .83 and .99 (as cited in Ayello, 2007, par. 4). Such aspects indicates that Braden scale has a favorable reliability for use in estimating risk of pressure ulcers in at-risk populations Go to conclusion.