Development of a hospital-acquired pressure injury predictive model using the Maryland Health Services Cost Review Commission Database (2012-2017)
Pragna N. Shetty, MPH1, Franca Kraenzlin, MD2, Justin M. Sacks, MD MBA FACS2.
1Johns Hopkins Bloomberg School of Public Health, Baltimore, MD, USA, 2Johns Hopkins Department of Plastic and Reconstructive Surgery, Baltimore, MD, USA.
Background: Hospital-acquired pressure injuries (HAPIs) cause significant morbidity and mortality to patients each year. In 2008, the Center for Medicare and Medicaid Services (CMS) declared them a “never event” to emphasize prevention. Despite this, it is estimated that they cause 60,000 deaths and cost between $9-11 billion dollars to treat annually. Risk factors for developing a HAPI have been identified, but how they interact with each other to contribute to HAPI development is not known. The purpose of this study was to develop a prediction model estimating risk of developing a stage 3 or stage 4 hospital-acquired pressure injury.
Methods: The Maryland Health Services Cost Review Commission database from 2012 to 2017 was used. Patients were identified from the medical, surgical, coronary care, and burn intensive care units. A total of 289,898 patients were found to meet inclusion criteria with 4,204 stage 3 and 4 HAPIs identified. A literature review was conducted to identify appropriate variables to include in this analysis. A univariate logistic regression model was used to conduct a risk factor analysis. This was used to determine which variables would be included in the final regression model. A 10-fold cross-validation and area under the curve test were conducted to observe the validation and accuracy of the model.
Results: The final predictive model showed that females were 0.92 times as likely to develop a HAPI as compared to male patients. Black patients were 2.01 times as likely to develop a HAPI as their white counterparts. Patients with diagnoses of spinal cord injury, diabetes mellitus, or stroke were also more likely to develop a HAPI than patients without these diagnoses. Both Medicare and Medicaid patients were at greater odds of developing a HAPI compared to the commercially insured cohort with ORs of 2.40 and 1.81, respectively. Those with body mass indexes classified as underweight had an OR of 2.39 of developing a HAPI compared to those who were normal or overweight. Those with morbid obesity were 28% less likely to develop a HAPI. The area under the curve for this model was 0.77, indicating good predictive accuracy.
Conclusions: This proposed prediction model has good accuracy at predicting which patients are more likely to develop a HAPI. While the tool should be externally validated on a different cohort of patients, it is a good starting point to help focus interventions and quality improvement for the prevention of this disease.
Back to 2019 Posters