Pressure Ulcer Risk Scales

Dear Readers:

Pressure ulcer risk scales (PURS) identify patients at risk of developing a pressure ulcer and guide prevention interventions for individual patients.¹ To function well in either capacity, a PURS should be reliable and valid. Inter-rater reliability is the capacity for different scale users to report the same risk for the same patient at the same time. Validity is usually measured after the study interval as the percent of patients that develop (sensitivity) or do not develop (specificity) a pressure ulcer (PU) matching the scale predictions. Predictive validity is the percent of patients enrolled in the study whom the scale predicted to develop a PU who actually did so (predictive value of a positive scale rating: PVP) or predicted to be PU-free who remained so (predictive value of a negative rating: PVN). Scale accuracy is the percent of both PU and PU-free predictions that were correct. Many studies have explored these metrics for PURS. This Evidence Corner reviews a recent meta-analysis of PURS validity and effectiveness in preventing PU, and a recent prospective comparison of PURS validity.

Valid, Reliable Pressure Ulcer Risk Assessment

Reference: Pancorbo-Hidalgo PL, Garcia-Fernandez FP, Lopez-Medina IM, Alvarez-Nieto C. Risk assessment scales for pressure ulcer prevention: a systematic review. J Adv Nurs. 2006;54(1):94–110.
Rationale: Pressure ulcers are a critical health care problem. There are several PURS potentially useful in PU prevention but not all have undergone proper validation.
Objective: Determine effectiveness of PURS degree of validation and use for PU prevention in clinical practice.
Methods: A systematic bibliographic review of literature on risk assessment and pressure or decubitus ulcer or sore summarized evidence from 14 databases. Efficacy studies and validity metrics were tabulated. A meta-analysis of validity metrics was also performed on the 4 PURS techniques with 2 or more valid original studies to determine the random effects model magnitude of the odds ratio of their capacity to predict PU development.
Results: The authors listed 3 studies investigating clinical effectiveness of the Norton or modified Norton Scale and 33 studies of PURS validation. No decrease in PU incidence was found that could be attributed directly to using a PURS, but PURS use increased intensity and effectiveness of prevention interventions. Only the Braden, Norton, and Waterlow Scales and clinical judgment had sufficient data for a meta-analysis comparing capacity for PU prediction. The Braden Scale showed optimal predictive efficacy (66.7% correct positive and negative predictions combined) and the best balance of sensitivity (57.1%) and specificity (67.5%). Sixteen studies on 5,847 patients supported a 4.08 odds ratio of correctly predicting a PU using the Braden Scale. The Norton Scale was the second most efficient predictor (60.2% correct), with 46.8% sensitivity, 61.8% specificity and an odds ratio of 2.16 of correctly predicting a PU based on 5 studies on 2,008 patients. Low specificity decreased predictive efficacy of the Waterlow Scale (34.4% correct) to below the 58.0% reported for clinical judgment, though the odds ratio of correctly predicting a PU in the meta-analysis was higher (2.05 based on 5 studies of 2,215 patients) than that of clinical judgment (1.69 based on 3 studies of 302 patients). Predictive value of a positive test was below 33% for all 4 techniques, possibly owing to generally low PU incidence and over prediction of PU as each PURS may have served to focus prevention interventions for patients identified to be at high risk for developing a PU.
Authors’ Conclusion: The Braden Scale has the best validity and reliability indicators across many studies and settings. Both Braden and Norton Scales predict PU development better than nurses’ clinical judgment, while the Waterlow Scale is highly sensitive but not very specific in predicting PU development. Use of a PU risk assessment scale improves PU preventive interventions but is not, by itself efficacious in decreasing PU incidence.

Predicting Pressure Ulcer Risk: A Prospective Study

Reference: Jalali R, Rezaie M. Predicting pressure ulcer risk: comparing the predictive validity of 4 scales. Adv Skin Wound Care. 2005;18(2):92–97.
Rationale: A first step in preventing pressure ulcers is determining which patients are at the highest risk of developing a PU.
Objective: Compare clinical validity and reliability of the 4 most commonly used PU risk assessment scales and identify a scale suitable for use in the hospital setting.
Methods: A prospective, nonrandomized study enrolled a convenience sample of 230 patients who were at least 21-years-old and PU-free on admission to the neurology, intensive care, orthopedic, and medical units of an Iranian hospital. Four independent research nurses administered the Braden, Norton, Gosnell, and Waterlow Scales to all patients within 48 hours of admission. All patients were assessed daily for development of a Stage I–IV PU for at least 14 days. It was determined that a score of 16 or greater for risk of developing a PU was the cut point for all 4 scales. Measures of predictive validity included accuracy, sensitivity, specificity, PVP, PVN, and the Youden Index of predictive power:
(Youden index = 1 – ([the probability of a false positive prediction] + [the probability of a false negative prediction]).
A Pearson product moment correlation was calculated between the number of subjects with a PU and the Youden score for each scale.
Results: Scale interrater reliability between the investigator and research staff was maintained at 0.95. At least 1 PU was noted in 74 of the 230 (32.2%) patients in the sample. Youden index, accuracy, sensitivity, specificity, PVP, and PVN of the tests are shown in the table below.
Authors’ Conclusion: The Gosnell scale was deemed most appropriate of the 4 scales for predicting PU risk in orthopedic and neurologic populations.

Clinical Perspective

Pancorbo-Hidalgo et al provide us with a quality review of PU validity and remind us that routine use of a PURS is merely a first step in PU management. Use each patient’s most recent PURS results to guide interventions to reduce PU incidence. It may seem confusing that a meta-analysis and a large prospective controlled study drew such different conclusions about PURS validity. Confusion may arise from use of different cut points for PU risk. For example, a Hong Kong study of 138 orthopedic or medical patients using a PU risk cut point of 18 instead of 16 reported 68% accuracy.² To correct for variations in patient acuity, comorbidities, facility case mix or interventions, the point of inflection of a receiver-operating curve plotting sensitivity as a function of specificity was often used in studies cited in the meta-analysis to determine the ideal PU risk cut point for each patient sample. This was not done in the prospective study reviewed. Also, it may be less confusing to view PU risk as the biologic continuum it is, rather than using a PURS as a dichotomy indicating if each patient is at risk or not at risk. Ayello and Braden¹ have taken a step toward the risk-continuum approach by interpreting a patient’s Braden Scale scores of 15–18 as “at risk,” 13–14 as “moderate risk,” 10–12 as “high risk,” and 9 or less as “very high risk.” Another difference between the 2 studies is the approach to evaluating predictive accuracy. Youden’s index is a respected epidemiologic tool, yet it is not clear that correlation with Youden’s index supports a conclusion of scale predictive power or its value to clinical users and their individual patients. Sensitivity and specificity describe percentages of patients with or without a PU who were accurately predicted by the scale at the end of the study. However, PURS are often used proactively as tools to predict and plan interventions to prevent a PU on individual patients. PVP and PVN may be more useful tools for such proactive individual use since they describe prospective predictive validity. For example, subscale examination can alert staff to attend to special patient needs like nutrition, pressure redistribution, or incontinence-related skin care interventions.³ Though such proactive interventions may decrease PURS validity measures by preventing the PU that the PURS was designed to predict, the resulting reduction in PU prevalence⁴ or incidence⁵ and improved economic outcomes are even more important clinical achievements than accurate PU prediction.