Level of inhospital care does not predict length of stay for patients admitted with congestive heart failure

Abstract

Study objectives: Risk stratification for congestive heart failure (CHF) has focused on the decision to admit. Once this decision is made, little guidance is provided about what level of care a specific patient might require. There are significant cost differences between ICU, telemetry, and routine care, and appropriate disposition is a necessity in this age of increasing health care costs. We studied a cohort of patients with an emergency department (ED) discharge diagnosis of CHF to determine the effect of level of care and other variables on length of stay (LOS). Our hypothesis was that the higher the level of care, the sicker the patient and the longer the expected LOS. Methods: This was a retrospective review of hospital discharges at an urban facility including all patients with a primary ED discharge diagnosis of CHF (<i>International Classification of Diseases, Ninth Revision</i> code 428). Age (10-year intervals), sex, ethnicity, comorbidities (secondary ED discharge diagnosis present in an aggregate ≥2%), and level of care were examined for their effect on LOS. Statistical significance was tested using Tukey's Studentized range test (HSD) on all main effects. A multivariate predictive model was then constructed. Results: There were 821 subjects in the cohort from July 2000 to June 2003 (59% men and 50% Latino). LOS was increased among Latinos (mean 6.3 days; 95% confidence interval [CI] 5.5 to 7.0) versus blacks (mean 4.0 days; 95% CI 3.5 to 4.5; <i>P</i>=.0036). Level of care also modified the LOS, with those patients admitted to the surgical ICU for procedures having the longest LOS (mean 17.8 days; 95% CI 10.5 to 25.0). Comparing medical ICU to telemetry to routine revealed no significant difference in LOS (6.2 versus 6.5 versus 5.2 days). Comorbidities also were associated with a varied LOS (range: asthma, 3.2 days to atrial fibrillation, 8.5 days). Neither age nor sex modified the LOS. The final model demonstrated that a prolonged LOS was predicted by ethnicity (parameter estimate 0.96; 95% CI 0.35 to 1.57), comorbidity (parameter estimate 0.15; 95% CI 0.02 to 0.28), and level of care (parameter estimate 0.73; 95% CI 0.50 to 0.97). For the overall model, <i>F</i> was equal to 10.50 and <i>P</i> value was less than .0001. Finally, we grouped the different level of care locations into ED care only, routine care, telemetry care, and ICU. Even with this crude grouping, there remained no difference in the LOS between routine and telemetry care (5.71 days versus 6.54 days), although the ICU did show a statistically longer LOS, as was to be expected. Conclusion: There was no contribution to this model of LOS by or medical ICU level of care. Once the decision to admit a patient with CHF was made, further guidance appeared to be necessary to risk stratify with regard to level of care needed while in the hospital.