Predicting emergency room visits and hospitalizations among hospice patients with cardiac disease

J Palliat Med. 2008 Oct;11(8):1142-50. doi: 10.1089/jpm.2008.0032.

Abstract

Background: One goal of quality palliative care in hospice is to limit emergency room visits and/or hospitalizations (ERVH).

Purpose: The purpose of this study was to determine predisposing factors that contribute to ERVH and devise a model to predict the probability of hospice cardiac disease patients having ERVH after hospice admission.

Methods: The study was a retrospective chart review of hospice cardiac patients comparing those with ERVH (n = 65) and those who died in their homes (n = 80). Data reduction strategy included bivariate tests and model-building analyses using logistic regression analysis.

Results: Some factors were associated with decreased odds of having ERVH: older patients (odds ratio [OR] = 0.94; p = 0.001), increased nursing visits (OR = 0.79; p = 0.001) and chaplain visits (OR = 0.48; p = 0.040), patients' medication compliance (OR = 0.27; p = 0.029), patients treated with morphine sulfate (OR = 0.15; p = 0.001), patients with caregiver(s) (OR = 0.09; p = 0.012), patients with hospice emergency kits (OR = 0.33; p = 0.004), interaction of the latter two variables (OR = 0.26; p = 0.001), patients with do-not-resuscitate orders (OR = 0.29; p = 0.001), Caucasian patients (OR = 0.22; p = 0.001), and the interaction of the latter two variables (OR = 0.22; p = 0.001). Other factors were associated with increased odds of having ERVH: patients with diabetes (OR = 2.64; p = 0.009), peripheral vascular disease (OR = 5.30; p = 0.003), hyperlipidemia (OR = 4.52; p = 0.013), chronic pulmonary disease/asthma (OR = 2.83; p = 0.003) as comorbidities; palpitations (OR = 6.61; p = 0.010), and chest pain (OR = 3.78; p = 0.006). The best fitting forecasting model had six independent variables decreasing and two increasing the odds of having ERVH. The final model chosen was: Z = 7.817 + 2.929 (peripheral vascular disease) + 1.513 (diabetes mellitus) - 3.306 (Caucasian) - 1.824 (caregiver presence x hospice emergency kit at home) - 0.212 (frequency of nursing visits) - 3.275 (on morphine) - 4.422 (medication compliance).

Conclusion: The forecasting model predicted the probability of ERVH correctly in 87.7% of the patients. The model is simple to use to predict hospice cardiac patients having ERVH. Future studies should validate this model. Interventions should utilize these factors and be evaluated to determine their ability to decrease ERVH in hospice cardiac patients.

MeSH terms

  • Attitude to Health
  • Emergency Service, Hospital*
  • Heart Diseases / therapy*
  • Hospice Care / methods*
  • Hospice Care / standards
  • Hospitalization*
  • Humans
  • Logistic Models
  • Models, Statistical*
  • Palliative Care / methods*
  • Palliative Care / standards
  • Probability
  • Quality-Adjusted Life Years
  • Retrospective Studies