Livingston Edward H, Ko Clifford Y
VAMC Greater Los Angeles Health Care System, California, USA.
J Surg Res. 2002 Jun 1;105(1):48-52. doi: 10.1006/jsre.2002.6448.
Multiple regression is the best technique for the simultaneous analysis of the contributions of multiple risk factors to a surgical outcome. A probability analysis is used to determine the relative contribution of individual factors to the overall outcome being assessed. We used these techniques to determine which of the potential risk factors had the greatest impact on adverse outcomes following gastric bypass surgery.
Records from 1067 consecutive patients undergoing Roux-Y gastric bypass at the UCLA Medical Center from December 1993 until June 2000 were reviewed. Major complications were used as the dependent variable in a multivariate logistic regression analysis, and 10 risk factors served as the independent variables. Based on the analysis, an average (i.e., baseline) patient was defined. Variations in the preoperative risk factors were then analyzed individually and in various combinations and their effect on the predicted probability for complication development was assessed.
The overall major complication rate in this series was 5.8%. The average patient was defined as a 334-pound woman who was 42.3 years of age. For this patient, the predicted complication rate by probability analysis was 3.9%. The greatest increase in the anticipated complication rate was attributable to revisional procedures that increased the rate to 6.5%. Many patients have a combination of risk factors; to this end, a 62-year-old, male patient with a 2SD increase in weight (i.e., 464 pounds) who was undergoing a revision operation and had a history of smoking, hypertension, diabetes, and sleep apnea had a predicted complication rate of 33.7%.
Probability analysis is a useful tool for determining the relative contribution of individual and combinations of risk factors for predicting the outcomes for surgical procedures. The four most influential factors for predicting a complication after gastric bypass surgery were; (1) male gender, (2) revisional surgery, (3) increasing age, and (4) increasing weight. These factors increased the predicted complication rate by 56, 67, 28 and 28%, respectively.
多元回归是同时分析多个风险因素对手术结果影响的最佳技术。概率分析用于确定各个因素对所评估的总体结果的相对贡献。我们运用这些技术来确定哪些潜在风险因素对胃旁路手术后的不良结果影响最大。
回顾了1993年12月至2000年6月在加州大学洛杉矶分校医学中心连续接受Roux - Y胃旁路手术的1067例患者的记录。在多因素逻辑回归分析中，将主要并发症作为因变量，10个风险因素作为自变量。基于该分析，定义了一个平均（即基线）患者。然后分别分析术前风险因素的变化及其各种组合，并评估它们对并发症发生预测概率的影响。
本系列患者的总体主要并发症发生率为5.8%。平均患者定义为一名体重334磅、年龄42.3岁的女性。对于该患者，通过概率分析预测的并发症发生率为3.9%。预期并发症发生率的最大增幅归因于翻修手术，该手术使发生率增至6.5%。许多患者存在多种风险因素；为此，一名62岁、体重增加2个标准差（即464磅）、正在接受翻修手术且有吸烟、高血压、糖尿病和睡眠呼吸暂停病史的男性患者，其预测并发症发生率为33.7%。
概率分析是确定个体及风险因素组合对手术结果预测相对贡献的有用工具。预测胃旁路手术后并发症的四个最具影响力的因素为：（1）男性；（2）翻修手术；（3）年龄增长；（4）体重增加。这些因素分别使预测并发症发生率提高了56%、67%、28%和28%。
