An optimization of the Waterlow score using regression and artificial neural networks

Evaluation of an Automated Pressure Ulcer Risk Assessment Model

The key to timely interventions and reducing avoidable incidence is the early identification of patients at risk for developing pressure ulcers. To enable the automatic detection of such patients and inform acute care interdisciplinary providers, a filter feature model using heuristic statistical methods was applied to a relational database of retrospective patient data including demographics, medications, and clinical visit details. These attributes served as input for the C4.5 decision tree induction algorithm, which was used to classify patient risk. The validity of the resulting classification model, Electronic Pressure Ulcer Prediction (ePUP), was assessed using a fourfold cross-validation. The current results show a limited application of such a naïve classification algorithm for automating pressure ulcer risk assessments. Additional refinements will be necessary before the predictions of ePUP are sufficient for general clinical use and the improvement of patient safety in acute care settings, and during the transition from hospital to home.

Investigating the reliability and validity of the waterlow risk assessment scale: a literature review

The aim of this review was to examine health literature on the reliability and validity of the Waterlow pressure sore assessment scale. A systematic review of published studies relating to the topic was conducted and literature was examined for its relevancy to the topic under investigation. Findings suggest that despite the availability of over 40 assessment tools, the Waterlow assessment scale is the most frequently used by health care staff. Research suggests that the Waterlow Scale is an unreliable method of assessing individuals at risk of pressure sore development with all studies indicating a poor interrater reliability status. Its validity has also been criticized because of its high-sensitivity but low-specificity levels.

Norton, Waterlow and Braden scores: a review of the literature and a comparison between the scores and clinical judgement

AIMS AND OBJECTIVES

To consider the validity and reliability of risk assessment scales for pressure ulcers.

BACKGROUND

Pressure ulcers are a major problem worldwide. They cause morbidity and lead to mortality. Risk assessment scales have been available for nearly 50 years, but there is insufficient evidence to state with any certainty that they are useful.

DESIGN

A literature review and commentary.

METHODS

Bibliographic databases were searched for relevant papers, a critical review was completed on relevant papers.

RESULTS

There is contradictory evidence concerning the validity of risk assesment scales. The interaction of education, clinical judgement and use of risk assessment scales has not been fully explored. It is not known which of these is most important, nor whether combining them results in better patient care.

CONCLUSIONS

There is a need for further work. A study exploring the complex interaction of risk assessment scales, clinical judgement and education and training is introduced.

RELEVANCE TO CLINICAL PRACTICE

Nurses may be wasting their time conducting risk assessment scoring if clinical judgement and/or education are sufficient to assess pressure ulcers risk.

Risk assessment scales for pressure ulcers: A methodological review

Much is written about risk-assessment scales (RASs) for pressure ulcers (PU) and their properties demonstrating that they are of limited value. Less is known about the reasons for these limitations and the scope for improvement. This review examines issues such as structure and scoring for the Norton, Waterlow and Braden scales, showing that the equal-weighting technique behind the current RASs is too simplistic and leads to limitations. It concludes that properly trained, experienced nurses should conduct PU risk assessments, whilst more robust data-driven RASs should be developed using the differential weighting scoring method together with advanced statistical techniques.

An evaluation of the Waterlow pressure ulcer risk-assessment tool

The National Institute for Clinical Excellence (NICE, 2001) stated that in order to achieve clinical governance strategies, risk management in the form of pressure ulcer risk assessment will contribute to improved quality of patient care. Risk-assessment tools must display high-predictive values, be highly sensitive and specific, reliable, and easy and convenient to use. Despite their extensively documented poor performance, pressure ulcer risk-assessment scales play a primary role in the prevention and management of pressure ulcers. The tools attempt to determine patients' risk status by quantifying a range of the most commonly recognized risk factors. Although there are over 40 different assessment tools, the Waterlow pressure ulcer risk-assessment tool is the tool that is most widely used in the UK. This article aims to review studies relating to the reliability, content validity, predictive validity, concurrent validity and the construct validity of the Waterlow pressure ulcer risk-assessment tool.

A regression analysis of the Waterlow score in pressure ulcer risk assessment

OBJECTIVE

To explore the predictive value of the Waterlow score, and the subscores of age and gender.

DESIGN

Logistic regression analysis was conducted on the two subscores of the Waterlow score, and the residual Waterlow score with gender and age removed. Receiver operating characteristic (ROC) analysis gave a quantitative measure of the classification ability of the Waterlow score.

SETTING

Burton, UK.

SUBJECTS

All admissions over a five-year period to the District General Hospital, a total of 150,015 admissions of 82,691 patients.

INTERVENTIONS

None.

MAIN OUTCOME MEASURES

Area under the ROC curve for significant (as determined by logistic regression) variables.

RESULTS

Data were inaccurate in at least 44.7% of the records, and analysis was conducted on the 43,735 records for which no errors were apparent. Nine hundred and fifty-four patients had a pressure ulcer on admission (2.1%); 277 developed a pressure ulcer (0.6%). The Waterlow score was predictive of pressure ulcers. Age was predictive, and gender was not found to be a significant predictor.

CONCLUSIONS

The Waterlow score may be improved and simplified by removing gender from the scoring system.

Improving the accuracy of pressure ulcer risk calculators: some preliminary evidence

This study presents data from a prospective cohort study of 213 in-patient admissions of people over 65. Logit analysis was used to investigate the relative contribution of a range of risk factors to the risk of pressure ulcer occurrence, as a basis for development of improved risk assessment tools. It was found that for this population, a model containing the Waterlow risk factors appetite, continence, skin condition and age, plus diagnosis, performed better than one based on the complete set of Waterlow factors. Gender was not significant. A diagnosis of cancer was positively associated with pressure ulcer occurrence but presence of Parkinson's disease had the opposite effect.

Classification of chronic radiation sickness cases using neural networks and classification trees

Chronic radiation sickness is a deterministic radiation health effect observed among the Mayak Production Association workers in Russia. In this study, unsupervised neural networks were used to cluster hematological measurements in a subset (n = 88) of the Mayak Production Association population while excluding from the analysis the radiation dose and the historical clinical diagnosis. Clusters of observations that had lower average leukocyte and thrombocyte counts were labeled "affected" and those having higher average blood cell counts were labeled "unaffected." The class (cluster) membership for each individual was used subsequently as a dependent variable in a classification tree model in order to identify significant features of the underlying classification model. After re-classification of cases using this method, the results showed a better data separation between the blood cell counts for affected vs. unaffected groups compared to those based on historical classification, and a greater difference between group means for differential blood counts was observed than for the historical diagnosis. The reclassification of diagnostic groups changed the group mean radiation doses. The geometric means (and 95% CL) of cumulative radiation dose equivalent from external exposures, based on the historical diagnosis, are 0.31 (0.0035, 3.4) vs. 1.7 (0.0007, 18) Sv. After clustering and classification tree analyses, the group geometric means were 0.78 (0.0014, 8.6) vs. 1.5 (0.0007, 17) and 0.82 (0.0013, 9.0) vs. 1.4 (0.0008, 16) Sv, using (respectively) whole blood cell counts or differential counts as the independent variables. The approach presented here is useful as a diagnostic aid for both retrospective analyses and in the event of future radiation accidents.

An investigation into the use of serum albumin in pressure sore prediction

OBJECTIVES

To ascertain the relevance of serum albumin and serum sodium as predictors of pressure sores in addition to the Waterlow score.

DESIGN

Observational study of patients at risk of developing decubitus ulcers.

SETTING

Staffordshire, in the midlands of the United Kingdom.

PARTICIPANTS

773 elderly hospital in-patients of a district general hospital.

MEASUREMENTS

Waterlow scores and serum albumin and sodium. Development of a pressure sore.

RESULTS

Logistic regression analysis of serum albumin, serum sodium and the Waterlow score showed the Waterlow score and serum albumin were significant predictors of pressure sores.

CONCLUSIONS

Serum albumin may, in this patient group (in-patients over 64 years of age), be a useful predictor of pressure sore occurrence, though further work is needed to establish whether this is the case. Risk assessment of pressure sores can possibly be improved by adding serum albumin to one of the pre-existing tools such as the Waterlow score.