PURPOSE: Studies have reported that patients transferred to higher levels of care following admission have excess mortality and length of stay (LOS). Additionally, it has been suggested that patients transferred to a tertiary care ICU are more severely ill and consume more resources.
METHODS: We performed a retrospective chart review of all patients admitted to the medical ICU at University of Kentucky (UKMC) between 2007 and 2009. Data was collected from hospital admissions data, US Census Bureau, and Social Security Death Index. Zip code was used for US Census data analysis of percentage of residents at the poverty level (PL) and percentage at two times the poverty level (PL2).
RESULTS: There were 2003 admissions to the ICU. Ten admissions were excluded due to lack of data. Mean age was 55.6±16.0 years; 88% were Caucasian decent and 54% were male. APACHE IV score ranged from 5 to 213 (median 69). The mean PL of the cohort was 20.6%±10.5% while the mean PL2 was 43.6%±14.7%. This compares to Kentucky rates of 15.8% and 35.9%, respectively. Multivariable regression analysis showed that age (OR 1.011; 95% CI 1.001-1.021) and APACHE IV score (OR 1.039; CI 1.035-1.044) were predictors of higher mortality. Race, gender, PL and PL2, insurance status, distance to UKMC, admitting source (ward vs. emergency department), use of dialysis, and day of admission were not associated with ICU mortality. ICU admissions from outside hospital transfers were associated with shorter LOS and lower mortality when compared to other admission sources (OR 0.610; CI 0.442-0.842). Similarly, total charges (142x103 vs. 197x103, p=0.0001) were different between outside hospital transfers and intra-hospital transfers.
CONCLUSIONS: Age and APACHE IV score are good predictors of mortality in our ICU. Compared to intra-hospital transfers, admissions from referral institutions are associated with shorter LOS, lower mortality and cost.
CLINICAL IMPLICATIONS: Inter-hospital transfer to an ICU is associated with lower mortality, cost and LOS. This may reflect referral bias. Further studies are needed to confirm these findings.
DISCLOSURE: The following authors have nothing to disclose: Enrique Diaz-Guzman, Rayan Ihle, Daniel Davenport, Ion Mitrache, David Mannino
No Product/Research Disclosure Information