Methodology: This is a two-phase project using retrospective case control data with cases (patients who developed HPU) and controls (patients who did not) selected at the ratio of 1:2, respectively. Research and literature review revealed 13 factors that appear to have influence on HPU development: Diabetes, Vascular , Neuropathy, Braden Scale score < 18, >70 years, immobility, perfusion issues, obesity, cachexia, surgery >3 hours, ICU stay >3 days, ventilator >3 days, and activity status.
For Phase 1, we have collected data on 150 patients with HPUs and 300 without. Currently, we are performing univariate analyses to identify patient characteristics and hospital conditions significantly associated with HPUs. These variables will be used in multiple logistic regression modeling. We will use ROC curves to select the most adequate model to predict HPUs based on these variables.
In Phase 2, the method will be validated using additional retrospective case-control data (30 with HPU, 70 without).
Statistics: Thus far, chi-square analysis was used to study the association of each variable with HPUs.
Results: Chi-square analysis shows that Vascular disease (OR=4.421, 95% CI, 2.573-7.595, X2=.000), Diabetes (OR=2.557, 95% CI, 1.652-3.958, X2=.000), and immobility (OR=16.491, 95% CI, 8.821-30.828, X2=.000) have a strong influence on HPU development. Neuropathy, >70 years, perfusion issues, obesity, cachexia, and ICU stay were not significantly related to HPU development. Analysis continues for Braden score, activity status, ventilator days, and surgical time.
Conclusion: Targeting certain hospitalized populations who are deemed more at-risk for HPUs, will allow implementation of appropriate prevention measures.