Investigating whether routinely collected biomarkers improve the prediction of hospital-acquired pressure injury occurrence: A retrospective cohort study

Background Despite being largely preventable, hospital-acquired pressure injuries remain a significant challenge in healthcare, contributing to prolonged hospital stays, increased patient morbidity, and substantial healthcare costs. Commonly used risk assessment tools have limited predictive accuracy, and early detection of hospital acquired pressure injuries often depends on subjective visual skin assessments. Emerging evidence suggests routinely collected biomarkers may offer an objective and reliable approach to predicting hospital acquired pressure injuries risk. Objective To explore how biomarkers improve hospital acquired pressure injuries prediction. Design Retrospective cohort study. Setting Acute NHS Trust in England, UK. Participants 10,504 adult patients admitted to acute medical wards for at least 24 h in 2024. Methods We considered the first hospital acquired pressure injuries and first record of other variables per hospital episode, measured at or soon after admission. Population characteristics of those who developed a hospital acquired pressure injuries or not were compared, overall and stratified by categories of length of stay (<6 days, ≥6 days), Braden, Malnutrition Universal Screening Tool and Mobility scores. Using multivariable logistic regression, we assessed the predictive value of the risk scores, adjusted for age and gender, and adding single biomarkers. Predictive performance was evaluated by discrimination and calibration. Analyses were exploratory. We used Stata v16 and R v4.4. Results Median hospital stay for patients with hospital acquired pressure injuries ( n = 293) was 18 days (interquartile range 12–31) compared with 5 days (2–11) for those without. Patients with hospital acquired pressure injuries were older than those without (84 (77–89) vs 78 (66–86) years. Levels of urea, C-reactive protein, and prothrombin time were significantly higher and albumin, haemoglobin and red blood cell count were significantly lower in those who developed hospital acquired pressure injuries. The incidence of hospital acquired pressure injuries was higher in those with longer hospital stays and increased across the risk score categories. Adjusting for age and gender, a unit increase in the Braden score reduced the odds of developing a hospital acquired pressure injuries by 15 %. The discrimination was adequate (AUC 0.72), but calibration was poor. Several individual biomarkers enhanced discrimination, but with miscalibration. Albumin was an independent predictor of hospital acquired pressure injuries in all models. The model with mobility adjusted for age and gender had adequate discrimination (AUC 0.71) and was well calibrated. Compared to those fully mobile, there was a sevenfold increase in the odds of hospital acquired pressure injuries in the immobile, and eightfold increase in those non-weight-bearing. Conclusions This study provides insight into the potential role of biomarkers in enhancing hospital acquired pressure injuries risk prediction. Further research should investigate how using objective biomarkers, combined with risk scores, improves the prediction of hospital acquired pressure injuries.