Issue:
A major pharmaceutical client sought to better understand which biomarkers associated with acute heart failure (AHF) were predictive of patient outcomes (readmission risk, mortality risk, lengthy length of stay and high costs).
Solution:
- We conducted a novel two stage GLM and logistic regression modeling analysis using baseline patient and hospital characteristics to determine which AHF-related biomarkers predicted the outcomes of interest.
- Using hospital claims data, we first identified key risk factors predictive of the outcomes of interest by constructing purposeful stepwise logistic regression and GLM models using patient-level and hospital-level baseline characteristics. The second stage was then to add the biomarker of interest to determine the impact of the biomarker on the outcome of interest, controlling for all other significant covariates. This process was repeated for each biomarker and for each outcome. The key outcomes indicated that there were a few CV and renal biomarkers that were consistently associated with outcomes. This finding has prompted the client to further explore the impact of these biomarkers in additional follow-up analyses.
- Two abstracts have been submitted to medical conferences and accepted for presentation. A manuscript highlighting the key findings from this analysis has been developed and will soon be submitted to a peer-reviewed scientific journal.
