Regional Validation and Updating of Clinical Predictive Models for Patients with Acute Heart Failure.
Background: It is increasingly common to incorporate risk-based analyses into clinical
decisions for patients with cardiovascular disease. While many current cardiovascular
guidelines, including those for heart failure, recommend integrating validated clinical
predictive models (CPMs), these predictive instruments are rarely used. One major
limitation is that heart failure is a ... read moreheterogeneous phenotype and the generalizability
of heart failure CPMs is largely unknown. Hypothesis: We hypothesize that regional CPMs
for patients with acute heart failure (AHF) do not generalize well to patients in other
world regions and that regional model performance can be improved with simple updating
procedures. Methods: We identified CPMs derived in North America that predict mortality
for patients with AHF and validated these models in different world regions to assess
performance in a contemporary international clinical trial (N = 4133) of patients with
AHF treated with guideline directed medical therapy. Results: We performed independent
external validations of 3 compatible CPMs predicting in-hospital mortality, 60 day
mortality, and 1 year mortality respectively. CPM performance varied substantially
across different world regions (North America, South America, Eastern Europe, and
Western Europe). The median percent decrease in discrimination across different world
regions was - 40% (range -15% to -92%). Regional calibration was highly variable
(Harrell's Eavg range <1% to >19%) however simple updating procedures
significantly improved local performance (recalibrated Eavg < 3% across all regions
and all models). Conclusions: Regional AHF CPM discrimination and calibration vary
substantially across different world regions and local performance must be understood
and optimized before these tools can be leveraged to improve patient
Thesis (M.S.)--Tufts University, 2016.
Submitted to the Dept. of Clinical & Translational Science.
Advisor: David Kent.
Committee: David Kent, Robin Ruthazer, and James Udelson.
Keyword: Medicine.read less
- Component ID:
- To Cite:
- TARC Citation Guide EndNote