OBJECTIVE: To facilitate the planning of perioperative care pathways, a fast-track failure prediction model has been developed in patients undergoing cardiac surgery. This study externally validated such a fast-track failure risk prediction model and determined the potential clinical consequences to ICU bed utilization. DESIGN: Prospective cohort study. SETTING: Cardiothoracic Surgery Department and Intensive Care Unit of Prince of Wales Hospital, The Chinese University of Hong Kong, Shatin, Hong Kong. PATIENTS: The St. Mary's Hospital fast-track failure risk prediction model was applied to patients included in an adult cardiac surgery database (January 2006 to June 2011). INTERVENTIONS: The performance of the fast-track failure risk model was assessed by discrimination and calibration methods. The potential clinical consequences of applying the model on ICU bed utilization was assessed using a decision curve analysis. MEASUREMENTS AND MAIN RESULTS: Of the 1,597 patients, 175 (11%) failed fast-track management. The final updated model showed very good discrimination (area under the receiver operating characteristic curve = 0.82, 95% confidence interval 0.78-0.86) and adequate calibration (Hosmer-Lemeshow goodness-of-fit statistic, p = 0.80). A decision curve analysis showed that if a threshold probability range of fast-track failure of 5% to 20% is used to determine who should be electively admitted to the ICU and who should be admitted to a fast-track recovery unit, it would lead to a substantial benefit (23%-67%) in terms of effective bed utilization, even after taking into account the negative consequences of unplanned admissions. CONCLUSIONS: As the performance of the final updated fast-track failure model was very good, it can be used to estimate the predicted probability of fast-track failure on individual patients. The clinical consequence of applying the final model appears substantial with regard to the potential increase in effective ICU bed utilization.
OBJECTIVE: To facilitate the planning of perioperative care pathways, a fast-track failure prediction model has been developed in patients undergoing cardiac surgery. This study externally validated such a fast-track failure risk prediction model and determined the potential clinical consequences to ICU bed utilization. DESIGN: Prospective cohort study. SETTING: Cardiothoracic Surgery Department and Intensive Care Unit of Prince of Wales Hospital, The Chinese University of Hong Kong, Shatin, Hong Kong. PATIENTS: The St. Mary's Hospital fast-track failure risk prediction model was applied to patients included in an adult cardiac surgery database (January 2006 to June 2011). INTERVENTIONS: The performance of the fast-track failure risk model was assessed by discrimination and calibration methods. The potential clinical consequences of applying the model on ICU bed utilization was assessed using a decision curve analysis. MEASUREMENTS AND MAIN RESULTS: Of the 1,597 patients, 175 (11%) failed fast-track management. The final updated model showed very good discrimination (area under the receiver operating characteristic curve = 0.82, 95% confidence interval 0.78-0.86) and adequate calibration (Hosmer-Lemeshow goodness-of-fit statistic, p = 0.80). A decision curve analysis showed that if a threshold probability range of fast-track failure of 5% to 20% is used to determine who should be electively admitted to the ICU and who should be admitted to a fast-track recovery unit, it would lead to a substantial benefit (23%-67%) in terms of effective bed utilization, even after taking into account the negative consequences of unplanned admissions. CONCLUSIONS: As the performance of the final updated fast-track failure model was very good, it can be used to estimate the predicted probability of fast-track failure on individual patients. The clinical consequence of applying the final model appears substantial with regard to the potential increase in effective ICU bed utilization.
Authors: Anna Lee; Chun Hung Chiu; Mui Wai Amy Cho; Charles David Gomersall; Kit Fai Lee; Yue Sun Cheung; Paul Bo San Lai Journal: BMJ Open Date: 2014-07-10 Impact factor: 2.692