Prediction: Readmission & Mortality After Discharge H. Tom Stelfox, Critical Care Canada Forum 2013
Disclosures No disclosures or conflicts of interest Shaun Hosein, MD, MSc.
Objectives 1. Provide a literature based estimate of readmission & death after patient discharge from ICU 2. Describe ICU discharge prediction rules 3. Present patient, provider & institutional factors associated with risk of readmission or death after patient discharge from ICU
Clinical Prediction tool that quantifies the individual contributions that various components of the history, physical examination, and laboratory results make towards the diagnosis, prognosis or likely response to treatment Derivation Identify factors with predictive power Validation Narrow Broad Impact Analysis Impact on behavior, outcomes, costs Laupacis JAMA 1997
Rationale for ICU Discharge Prediction Transitions of care are common & associated with adverse events ICU discharge is reported to be high risk Vulnerable patients resources Non-standardized Poor communication Multiple professions & specialties
Methods Search Strategy Databases: Medline, EMBASE, CINAHL, Cochrane, PubMed Search terms: Additional searches: Hand searched select journals Bibliography review Experts in field Adverse Event ICU Discharge
Methods Selection Strategy Inclusion criteria: Original research Adult patients discharged from ICU Evaluated discharge: risk factors or risk stratification tool Adverse event = MET, readmission, mortality Exclusion criteria: Coronary care units (CCU) High dependency units & step-downs Pediatric patients 2 authors independently reviewed all publications: Rater agreement very good (kappa 0.84)
Study Flow 7,923 Titles & Abstracts 148 Full Text Articles 58 Articles Included
Studies Included in Review Country USA, Australia, UK, Canada, Germany ICU Both single & multiple mixed medical-surgical ICUs Patients Total patients > 2 million Typical patient: 60yrs, male, APACHE II score 15-20 Methodology Cohort studies with multivariable modeling
What is the rate of readmission or death after patient discharge from ICU?
Readmission after ICU Discharge 6.3% (95% CI 5.5-7.2%)
Death after ICU Discharge 7.4% (95% CI 6.6-8.2%)
Potential Reasons for Variation Factor Readmission Hospital Mortality CVICU vs. Med-Surg ICU Age >60 vs. <60 yrs. Severity of Illness DNR patients included
What ICU discharge prediction rules have been developed?
Prediction Rules Tool Outcome Variables Sensitivity Specificity ROC Swift Score Readmit <7d 5 27% 87% 0.70 Frost et al. Readmit 7 NA NA 0.66 Reini et al. Readmit <72h 5 15% 85% NA Badawi et al. Readmit <48h 23 6-96% 19-99% 0.71 Death <48h 26 47-82% 87-99% 0.92 Daly et al. Hosp death 5 74% 71% 0.80 Sabadell Score Hosp death Subjective 26-85% 71-99% 0.84 MIR Score Readmit/death <7d 5 50-96% 19-82% 0.74 Hosein Crit Care 2013 MIR Score vs. SWIFT Score - ROC 0.74 vs. 0.61
What patient, provider & institutional factors are associated with risk of readmission or death after patient discharge from ICU?
Risk Factors for ICU Discharge
Risk Factors for ICU Discharge Age Male sex Chronic respiratory disease Chronic liver disease Severity of illness Length of ICU stay
What To Do With The Information? Help inform provider decision making but the challenge is that many of the risk factors are not modifiable For high risk patients Early initiation of discharge planning Allow more time for physiology to improve Use high dependency units to facilitate transitions Target follow up services For low risk patients Efficiently get them out of the ICU Quality assurance set performance targets & risk adjust outcomes
Potential role for Outreach Services Niven Crit Care Med 2013
Key Findings Risk of readmission & death Varies between studies 6% to 7% Prediction rules exist & there is a need for Comparative evaluation Evaluation of implementation Risk stratification may be possible, but unknown if it will improve patient care
Conclusions Individual risk factors to consider based on current evidence: Age Male sex Chronic respiratory disease Chronic liver disease Severity of illness Length of ICU stay
Acknowledgements Mentors Sharon Straus Bill Ghali Research Team Shaun Hosein Nik Bobrovitz Simon Berthelot Funding Agencies CIHR Alberta Innovates