Who Would Benefit From a Palliative Care Consult in the Hospital? CARING About Prognosis (CAP) (313-C)


      • 1.
        Identify hospitalized patients at high risk of death in one year.
      • 2.
        Apply the CARING criteria to patients on hospital admission.
      Background. The CARING criteria, which accurately predict death at 1 year among hospitalized veterans, have not been validated in other patient populations. Identifying patients at high risk for 1-year mortality at the time of hospital admission provides clinicians with a unique opportunity to consider palliative care interventions for these patients during hospitalization.
      Research objectives. Validate the CARING criteria among general medical and surgical patients at a tertiary academic center.
      Methods. Retrospective cohort study of all admissions to the general medicine and surgical services of the University of Colorado Hospital from July 2005 through August 2005 (n = 427). Medical records were reviewed for demographic information and presence of the CARING criteria (C = primary diagnosis of cancer, A = Admit more than two times in the past year for a chronic medical condition, R = resident in a nursing home, I = ICU with multi-organ failure, NG = NHPCO non-cancer hospice guidelines). Only information available at the time of admission was used to determine the CARING criteria. Mortality data came from vital records. Logistic regression models were fit for 1-year mortality predicted by the four dichotomous CARING variables (excluding ICU) and age by quartile.
      Results. One hundred and sixty-one (37.7%) patients met at least one of the CARING criteria and 77 (18%) patients died during the 1-year follow-up period. The validation sample differed from the original CARING sample with more females (50% versus 2%) and younger age (mean age 54 versus 63 years). Logistic regression odds ratios with 95% CI: Cancer = 4.0 (CI, 2.2-7.4); Admit = 1.3 (CI, 0.7-2.4); Resident = 1.6 (CI, 0.5-5.0); NHPCO guidelines = 6.0 (CI, 2.3–15.9). These findings are quantitatively similar to the original CARING logistic regression model.
      Conclusion. The CARING criteria maintain statistical validity in a more broadly representative patient population.
      Implications for research, policy, or practice. The CARING criteria should be considered as a simple mnemonic tool to identify patients on hospital admission who may benefit from palliative care.


      Structure and Processes of Care