(205) There's something about Albumin: Applying the Roadmap for Causal Inference to Estimate the Effect of Albumin Levels on Pressure Injury Development in Hospital Settings

Background: Pressure injuries (PrIs) are a common problem in critical care settings, affecting 3-6% of patients. These injuries are characterized by damage to the skin and underlying soft tissue. While external pressure and shear are the primary causes of PrIs, other factors can also contribute to tissue damage. One such factor is serum albumin, a protein found in blood plasma. Studies have shown that low serum albumin levels can increase the risk of developing PrIs.

Objectives: We aimed to apply the "roadmap for causal inference" to evaluate the impact of low albumin levels on the incidence of pressure injuries in hospital patients while considering any potential confounding factors that could be present in the population being studied.

Methods: Our study employed a plausible directed acyclic graph (DAG) relating albumin to pressure injuries with potential confounders, colliders, and effect modifiers (moderators). We used a nonparametric causal structural framework and employed the roadmap for causal inference and the Targeted Maximum Likelihood Estimation (TMLE) approach to adjust for confounders and estimate the causal effect of low albumin on the development of pressure injuries. Our analysis involved a cohort of patients identified from the MIMIC IV dataset, a de-identified electronic health records dataset from Beth Israel Deaconess Medical Center.

Results: Our analysis showed that low albumin levels are significantly associated with an increased incidence of pressure injuries in hospitalized patients, even after accounting for various other factors. The crude odds ratio for low albumin was strikingly high at 3.97 (95% confidence interval: 3.54-4.45). We used TMLE to refine these estimates, and we found that low albumin levels increase the risk of pressure injury by 4.0% (95% confidence interval: 0.03082-0.0496, p < 0.0001). We also discovered a dose-response relationship between albumin levels and pressure injury risk, further supporting the association.

Conclusions: Utilizing advanced causal inference techniques, this study provides robust evidence of the contributory role of serum albumin in the development of pressure injuries (PrIs) in critical care. Despite observing a significant variation in PrI risk among patients with liver and kidney diseases, our results did not indicate a significant effect modification, likely due to the increased variability in these sicker patient populations. Nevertheless, the evidence points to low albumin as a potential target for intervention. Our findings lay the groundwork for designing an interventional study with a systematic approach to albumin monitoring and intervention, which could reduce PrI incidence.