Predictors of early postdischarge mortality in critically ill patients: A retrospective cohort study from the California Intensive Care Outcomes project

Philip R. Lee Institute for Health Policy Studies, University of California at San Francisco, San Francisco, CA, USA.
Journal of critical care (Impact Factor: 2.19). 02/2011; 26(1):65-75. DOI: 10.1016/j.jcrc.2010.06.010
Source: PubMed

ABSTRACT Existing intensive care unit (ICU) mortality measurement systems address in-hospital mortality only. However, early postdischarge mortality contributes significantly to overall 30-day mortality. Factors associated with early postdischarge mortality are unknown.
We performed a retrospective study of 8484 ICU patients. Our primary outcome was early postdischarge mortality: death after hospital discharge and 30 days or less from ICU admission. Cox regression models assessed the association between patient, hospital, and utilization factors and the primary outcome.
In multivariate analyses, the hazard for early postdischarge mortality increased with rising severity of illness and decreased with full-code status (hazard ratio [HR], 0.33; 95% confidence interval [CI], 0.21-0.49). Compared with discharges home, early postdischarge mortality was highest for acute care transfers (HR, 3.18; 95% CI, 2.45-4.12). Finally, patients with very short ICU length of stay (<1 day) had greater early postdischarge mortality (HR, 1.86; 95% CI; 1.32-2.61) than those with longest stays (≥7 days).
Early postdischarge mortality is associated with patient preferences (full-code status) and decisions regarding timing and location of discharge. These findings have important implications for anyone attempting to measure or improve ICU performance and who rely on in-hospital mortality measures to do so.

Download full-text


Available from: Michael W Kuzniewicz, Apr 29, 2015
  • [Show abstract] [Hide abstract]
    ABSTRACT: Hospitals vary widely in the services they offer to care for pediatric burn patients. When a hospital does not have the ability or capacity to handle a pediatric burn, the decision often is made to transfer the patient to another short-term hospital. Transfers may be based on available specialty coverage for children; which adult and non-teaching hospitals may not have available. The effect these transfers have on costs and length of stay (LOS) has on pediatric burn patients is not well established and is warranted given the prominent view that pediatric hospitals are inefficient or more costly. The authors examined inpatient admissions for pediatric burn patients in 2003, 2006, and 2009 using the Kids' Inpatient Database, which is part of the Healthcare Cost and Utilization Project. ICD-9-CM codes 940 to 947 were used to define burn injury. The authors tested if transfer status was associated with LOS and total charges for pediatric burn patients, while adjusting for traditional risk factors (eg, age, TBSA, insurance status, type of hospital [pediatric vs adult; teaching vs nonteaching]) by using generalized linear mixed-effects modeling. A total of n = 28,777 children had a burn injury. Transfer status (P < .001) and TBSA (P < .001) was independently associated with LOS, while age, insurance status, and type of hospital were not associated with LOS. Similarly, transfer status (P < .001) and TBSA (P < .001) was independently associated with total charges, while age, insurance status, and type of hospital were not associated with total charges. In addition, the data suggest that the more severe pediatric burn patients are being transferred from adult and non-teaching hospitals to pediatric and teaching hospitals, which may explain the increased costs and LOS seen at pediatric hospitals. Larger more severe burns are being transferred to pediatric hospitals with the ability or capacity to handle these conditions in the pediatric population, which has a dramatic impact on costs and LOS. As a result, unadjusted, pediatric hospitals are seen as being inefficient in treating pediatric burns. However, since pediatric hospitals see more severe cases, after adjustment, type of hospital did not influence costs and LOS. TBSA and transfer status were the predictors studied that independently affect costs and LOS.
    Journal of burn care & research: official publication of the American Burn Association 12/2014; 36(1). DOI:10.1097/BCR.0000000000000206 · 1.55 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: Existing risk adjustment models for intensive care unit (ICU) outcomes rely on manual abstraction of patient-level predictors from medical charts. Developing an automated method for abstracting these data from free text might reduce cost and data collection times. To develop a support vector machine (SVM) classifier capable of identifying a range of procedures and diagnoses in ICU clinical notes for use in risk adjustment. We selected notes from 2001-2008 for 4191 neonatal ICU (NICU) and 2198 adult ICU patients from the MIMIC-II database from the Beth Israel Deaconess Medical Center. Using these notes, we developed an implementation of the SVM classifier to identify procedures (mechanical ventilation and phototherapy in NICU notes) and diagnoses (jaundice in NICU and intracranial hemorrhage (ICH) in adult ICU). On the jaundice classification task, we also compared classifier performance using n-gram features to unigrams with application of a negation algorithm (NegEx). Our classifier accurately identified mechanical ventilation (accuracy=0.982, F1=0.954) and phototherapy use (accuracy=0.940, F1=0.912), as well as jaundice (accuracy=0.898, F1=0.884) and ICH diagnoses (accuracy=0.938, F1=0.943). Including bigram features improved performance on the jaundice (accuracy=0.898 vs 0.865) and ICH (0.938 vs 0.927) tasks, and outperformed NegEx-derived unigram features (accuracy=0.898 vs 0.863) on the jaundice task. Overall, a classifier using n-gram support vectors displayed excellent performance characteristics. The classifier generalizes to diverse patient populations, diagnoses, and procedures. SVM-based classifiers can accurately identify procedure status and diagnoses among ICU patients, and including n-gram features improves performance, compared to existing methods.
    Journal of the American Medical Informatics Association 04/2014; 21(5). DOI:10.1136/amiajnl-2014-002694 · 3.93 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: Performance assessments based on in-hospital mortality for ICU patients can be affected by discharge practices such that differences in mortality may reflect variation in discharge patterns rather than quality of care. Time-specific mortality rates, such as 30-day mortality, are preferred but are harder to measure. The degree to which the difference between 30-day and in-hospital ICU mortality rates-or "discharge bias"-varies by hospital type is unknown. The aim of this study was to quantify variation in discharge bias across hospitals and determine the hospital characteristics associated with greater discharge bias. Retrospective cohort study. Nonfederal Pennsylvania hospital discharges in 2008. Eligible patients were 18 years old or older and admitted to an ICU. None. We used logistic regression with hospital-level random effects to calculate hospital-specific risk-adjusted 30-day and in-hospital mortality rates. We then calculated discharge bias, defined as the difference between 30-day and in-hospital mortality rates, and used multivariable linear regression to compare discharge bias across hospital types. A total of 43,830 patients and 134 hospitals were included in the analysis. Mean (SD) risk-adjusted hospital-specific in-hospital and 30-day ICU mortality rates were 9.6% (1.3) and 12.7% (1.5), respectively. Hospital-specific discharge biases ranged from -1.3% to 6.6%. Discharge bias was smaller in large hospitals compared with small hospitals, making large hospitals appear comparatively worse from a benchmarking standpoint when using in-hospital mortality instead of 30-day mortality. Discharge practices bias in-hospital ICU mortality measures in a way that disadvantages large hospitals. Accounting for discharge bias will prevent these hospitals from being unfairly disadvantaged in public reporting and pay-for-performance.
    Critical care medicine 01/2014; 42(5). DOI:10.1097/CCM.0000000000000132 · 6.15 Impact Factor