摘要
目的 急性心肌梗死患者直接PCI术后合并肠梗阻的预测因素,并建立预测模型.方法 回顾分析2004年6月至2014年6月期间行直接PCI术的患者共计1220例,其中符合入选、排除标准的1 025例,将入选的1 025例患者用SPSS17.0软件产生随机数字并随机分为2个队列,其中排序前2/3为模型产生队列,后1/3为模型验证队列.根据模型产生队列是否发生肠梗阻又分成肠梗阻组和对照组并分析肠梗阻的临床特点.将肠梗阻组与对照组比较后,差异具有统计学意义的参数纳入单因素Logistic回归分析,对有统计学意义的参数再进一步行多因素Logistic回归分析,从而确定预测因素.按β回归系数所占的比重,对预测因素给予赋值,建立评分模型,采用ROC曲线在模型验证队列中评价评分模型的预测能力.结果 1 025例患者中共有103例患者发生不同类型肠梗阻,发生率约10.0%,肠梗阻类型以低位、不全性及单纯性肠梗阻为主.多因素Logistic回归分析显示,年龄> 65岁(OR=1.44,95%CI:1.26~4.63,P=0.000)、糖尿病(OR=3.37,95% CI:2.39~9.47,P=0.000)、便秘习惯(OR=4.75,95% CI:3.58 ~ 11.24,P=0.024)、下壁和广泛前壁部位心梗(OR=2.16,95% CI:1.94-4.79,P=0.017)、心功能≥3级(killip分级)(OR=2.86,95% CI:1.98 ~ 5.67,P=0.002)、经股动脉穿刺(OR=2.76,95% CI:1.38 ~ 6.12,P=0.002)、K+<3.5 mmol/L(OR=1.86,95%CI:1.11 ~ 5.47,P=0.005)、通便药(OR=3.59,95% CI:2.99~10.21,P=0.000)及吗啡使用(OR=1.98,95% CI:1.07-3.12,P=0.021)及eGFR <60 mL·min-1.1.73m-2(OR=1.19,95%CI:1.10 ~3.22,P=0.031)是AMI患者直接PCI术后并发肠梗阻的预测因素.受试者工作特征(ROC曲线)证实根据这些因素建立的评分模型对AMI患者直接PCI术后肠梗阻的发生具有较好的预测准确性(AUC =0.815,95% CI:0.802~0.883,P=0.000).结论 年龄>65岁、糖尿病、便秘习惯、下壁和广泛前壁部位心梗、心功能≥3级(killip分级)、经股动脉穿刺、K+<3.5 mmol/L、通便药及吗啡使用及eGFR <60 mL· min-1·1.73m-2是AMI患者直接PCI术后并发肠梗阻的预测因素.由此建立的评分模型对直接PCI术后发生肠梗阻具有良好的预测能力.
Objective to determine predictive factors of intestinal obstruction in acute myocardial infraction (AMI) patients after primary percutaneous coronary intervention (PCI) and to establish predictive model.Methods A total of 1220 AMI patients underwent primary PCI in Jiangxi provincial people's hospitalfrom June2004 toJune 2014were retrospectively analyzed.And 1025 cases of them whichall met the inclusion and exclusion criteriawere randomlydivided by using random number generated by SPSS 17.0 into two cohorts:model derivation cohort (MDC) and model validation cohort (MVC).MDC was divided into intestinal obstruction group and control group.predictive factors were identified using univariable andmultivariable logistic regression analysis in MDC.Interger point values were assigned to each predictor based upon their coefficient in multivariable logistic regression model to establish scoring model.The summed scores of each case in MVC were calculated to test predictive ability of the model by ROC cure.Results Total of 1 025 patients,103 patients were diagnosed with different types of intestinal obstruction and the incidence of intestinal obstruction was 10.0%.Low position intestinal obstruction,incomplete intestinal obstruction and simple intestinal obstructionwere main characteristics in these patients.logistic regression analysis identified ten risk factorscan predict the appearance of intestinal obstruction:age 〉 65years old (OR =1.44,95% CI:1.26-4.63,P =0.000),diabetes mellitus (OR =3.37,95% CI:2.39-9.47,P =0.000),habitual constipation (OR =4.75,95 % CI:3.58-11.24,P =0.024),inferior and extensive anterior wall myocardial infarction (OR =2.16,95% CI:1.94-4.79,P =0.017),cardiac functiongrade≥3 class (Killip classification) (OR=2.86,95%CI:1.98-5.67,P=0.002),femoral approach (OR=2.76,95%CI:1.38-6.12,P=0.002),K+ 〈3.5 mmol/L (OR=1.86,95%CI:1.11-5.47,P =0.005),taking laxative (OR =3.59,95% CI:2.99-10.21,P =0.000),using morphine (OR =1.98,95% CI:1.07-3.12,P =0.021) and estimated glomerular filtration rate (eGFR) 〈60 mL · min-1 · 1.73m-2 (OR =1.19,95% CI:1.10-3.22,P =0.031).and areas under the ROC cure was 0.815 (95% CI:0.802-0.883,P =0.000) and indicating the accuracy of predicting intestinal obstruction.Conclusions age 〉 65 years old,diabetes mellitus,habitual constipation,inferior and extensive anterior parts of myocardial infarction,cardiac functiongrade ≥3class (Killip classification),femoral approach,K + 〈3.5mmol/L,taking laxative,using morphine and eGFR 〈 60 ml.min-1.1.73m2 were important predictive factors for intestinal obstruction in AMI patients after underwent 0 and the scoring model can predict accurately intestinal obstruction in such patients.
出处
《中华急诊医学杂志》
CAS
CSCD
北大核心
2015年第10期1164-1170,共7页
Chinese Journal of Emergency Medicine
关键词
急性心肌梗死
直接PCI
肠梗阻
预测
Acute myocardial infraction
Primary percutaneous coronary intervention
Intestinalobstruction
Forecasting