A Prognostic Indicator for Patients Hospitalized with Heart Failure

Area of Science:

• Health Services Research
• Gerontology
• Clinical Informatics

Background:

• Current methods for identifying patients nearing end-of-life are prone to clinician bias, leading to underestimation of mortality risk.
• This results in delayed access to essential hospice and palliative care services for many patients.

Purpose of the Study:

• To develop a prospective prognostic model using Centers for Medicare and Medicaid Services (CMS) claims data.
• The model aims to identify patients with congestive heart failure (CHF) at high risk for six-month mortality.

Main Methods:

• Analysis of CMS claims data from January 1, 2008, to June 30, 2009.
• Inclusion of CHF patients with diagnosis-related groups (DRGs) 291, 292, and 293.
• Univariate and multivariable analyses to identify predictors of six-month mortality, followed by model evaluation for discrimination and calibration.

Main Results:

• The developed prognostic model demonstrated fair discrimination (ROC 0.71) and good calibration (Hosmer-Lemshow statistic 0.98).
• Approximately 5% of discharged CHF patients across all DRGs exhibited a greater than 50% risk of mortality within six months.

Conclusions:

• This prospective claims-data-driven approach can identify CHF patients who may benefit from hospice or palliative care consultations.
• The model offers a more consistent method for matching at-risk patients with appropriate care.
• Further prospective evaluation is needed to confirm clinical utility, impact on referrals, and patient/family benefits.