|
|
October 16-20, 2010 San Diego, CA Home Abstract Archive Search Abstracts 2010 Session Grid 2010 Meeting Website Policy Statements ASA Website Feedback 
Visit Anesthesiology.org |
| Previous Abstract | Next Abstract |
| Printable Version |
|
A973
October 20, 2008 9:00 AM - 11:00 AM Room Hall E2-Area D, |
|||||||||||||||||||||||||||||||||||
| Derivation and Validation of a Predictive Model of Transfusion in Spine Surgery (PMTSS) in Adults. | |||||||||||||||||||||||||||||||||||
|
Paul Merckx, M.D., M.B.A., Brigitte Lenoir, M.D., Catherine Paugam-Burtz, M.D., Pierre Guigui, M.D., Ph.D., Jean Mantz, M.D., Ph.D. Department of Anesthesia and Critical Care, Beaujon University Hospital, AP-HP, Clichy, France |
|||||||||||||||||||||||||||||||||||
|
Allogeneic red blood cells (RBC) requirements in major thoracolumbar adult spine surgery remain important. The purpose of this work was to identify factors associated with perioperative transfusion of homologous RBC in patients undergoing elective surgery and to work out a consistent model of prediction. Methods: 1. Derivation retrospective study: 230 independent observations over a 15-month period in a unique tertiary care centre were included in the sample. Exclusion criteria were exclusive cervical spine surgery, one level laminectomy and polytrauma. Independent predictive factors of allogeneic RBC transfusion from the day of surgery until postoperative day 5 were identified using bivariate statistical analysis, recursive partitioning and multiple logistic regression modeling (JMP™V.7, SAS Institute). P<0.05 was considered significant. A simplified score-based model was designed using adjusted Odds Ratios (Ors) from model estimates and tested. The score discrimination and optimal cut-point were assessed using a receiver operating characteristic (ROC) curve. 2. Validation prospective study: Calibration and discrimination of the score were checked in a second predefined-size sample of 125 consecutive patients over a 6-month period in the same centre (narrow validation). Results: 1. Derivation sample: Homologous RBC transfusion (T) rate was 32%; when (T), RBC (units) median=3.5, 25%-75% range [2, 5]. Four independent variables were identified in a reliable model (Goodness-of-Fit (GOF) test X2=12, d.f.=15, p=0.67) as predictive factors: age (yrs), preoperative hemoglobin (Hb) (g.dl-1), spine fusion levels (n), and transpedicular osteotomy. An individual score (0 to 4 points) was then calculated. The maximum amount of 4pts was allocated in case of osteotomy, otherwise a total score was obtained with the sum of 1pt if spine fusion levels>2, 2pts if Hb<12, 1pt if 12≤Hb≤14, and 1pt if age>50. Using this score to predict a transfusion (GOF X2=1.3, d.f.=3, p=0.72), area under ROC curve (AUROC)=0.87, 95% confidence interval (CI) [0.81, 0.92]. When the score>2, (pretest p(T)=0.32) positive predictive value (PPV)=0.74, predictive negative error=0.12. Moreover, the score level was clearly associated with the count of RBC units. 2. Validation sample: (T) rate was 37%. Observed transfusion frequencies were not statistically different from expected values (GOF X2=4.1, d.f.=2, p=0.12;). Considering PPVs with transfusion rates at each possible score cut-point confirmed an adequate fit of the model. Discrimination was effective, AUROC=0.84, 95% CI [0.76, 0.92].[table1][table2]Conclusion: A simplified transfusion prediction score was derived then validated. Evaluation of its impact to manage blood requirements is the next step. A broad validation could proceed and involve different institutions. Anesthesiology 2008; 109 A973 |
|||||||||||||||||||||||||||||||||||
|
|||||||||||||||||||||||||||||||||||
|