The Braden Scale and Care Dependency Scale each demonstrate at least 70% sensitivity and specificity for identifying inpatients at risk of pressure ulcer
- Correspondence to Jürgen Stausberg
Institut für Medizinische Informationsverarbeitung, Biometrie und Epidemiologie (IBE), Ludwig-Maximilians-Universität München, Marchioninistraße 15, 81377 München, Germany;
The assessment of the patient's individual risk for pressure ulcers (PU) is a prerequisite for an adequate application of preventive interventions. For example, the new guideline of National Pressure Ulcer Advisory Panel and European Pressure Ulcer Advisory Panel recommends several preventive interventions solely for patients ‘at risk’.1 Therefore, the correct implementation of this guideline demands the differentiation between ‘at risk’ and ‘no risk’. However, the same guideline reports only indirect evidence for the use of a structured approach of risk assessment. Furthermore, no specific Risk Assessment Scale (RAS) is recommended. In case of an optimal RAS, each patient ‘at risk’ will develop a PU if he/she does not receive any preventive interventions and his/her conditions do not change. The proportion of positive predictions is called ‘positive predictive value’ (PPV). Each patient with ‘no risk’ will not develop a PU if his/her conditions do not change. The proportion of correct negative predictions is called ‘negative predictive value’ (NPV). PPV and NPV define the practical relevance of a RAS.
Concordance and diagnostic accuracy
Tannen and colleagues measured the concordance of two specific RAS (Braden and Waterlow) and one generic assessment tool (Care Dependency Scale, CDC) on the one hand and compared their diagnostic quality on the other hand. The study includes 1053 inpatients of one university clinic in Germany. Trained nurses were responsible for the recording of the scales and the inspection of skin. CDC and Braden scale showed the highest concordance. The latter succeeded in diagnostic quality with CDC close behind. The superiority of the Braden scale corresponds with the literature.2 Moreover, the study confirms the idea of using generic instruments for the assessment of PU risk.3
Unfortunately, neither PPV nor NPV is given in the article. PU prevalence – imperative for the calculation of PPV and NPV – is not mentioned either. Considering two different prevalences in German hospitals, 2%4 and 11%,5 the following results are achieved with the Braden scale. Ninety-four per cent or 71% of the patients ‘at risk’ will never develop a PU, even without any preventive intervention; 0.4% and 2.4% (in comparison to 2% and 11% of all patients) of the patients with ‘no risk’ will develop a PU. Thus, an uncontrolled use of the Braden scale will lead to a waste of resources and establish the risk for adverse events related to preventive interventions like repositioning. This holds true for the CDC applied to PU risk assessment.
Relevant information on both study design and study results is missing. For example, nothing is said about the risk assessment regularly applied in the study centre. It will introduce bias if one of the scales is already in use. Another limitation is inherent to the standard study design used for the evaluation of a RAS. It is ethically not permissible to omit preventive interventions in case of risk. Therefore, it is proposed to use the implementation of preventive measures as additional end point and to use the implementation of preventive measures.6 Tannen and colleagues applied a slightly different approach. They do not count the development of PU after risk assessment but rather the simultaneously existence of a PU at the time of risk assessment. It should be rethought whether diagnostic accuracy is a good proxy for prognostic quality.
Assessment of PU risk remains dissatisfactory even after reading this article. The available RAS are no more than an essential complement of nursing expertise. Therefore, it is not possible to link results of RAS directly to preventive interventions in an algorithmic way. To go one step back to a generic assessment of patients' care, as the authors recommend it, is a valuable conclusion that reduces workload in daily nursing practice. Future research about RAS for PU should concentrate on methodological issues, for example, on the appropriate trial design that reduces bias and counts valid outcome parameters.