Application of Logistic Regression and Decision Tree Model in Prediction of Adverse Reactions of Iodinated Contrast Medium
-
摘要:
目的 采用Logistic回归与决策树模型分析碘造影剂(iodinated contrast medium,ICM)不良反应影响因素。 方法 不良反应组为2019年1月至2023年8月期间在云南省阜外心血管病医院行CT检查使用ICM并出现不良反应的患者129例,对照组为同期使用ICM未发生不良反应的患者135例,将单因素分析中有统计学意义的变量纳入多因素Logistic回归和决策树模型,提取引起不良反应的影响因素并进行分析。 结果 Logistic回归结果显示BMI、过敏史和肾功能不全是发生ICM不良反应的影响因素(P < 0.05),预测模型为Lggit(P) = 0.123×BMI+1.684×过敏史+2.551×肾功能不全−3.652;决策树分析筛选出3个风险变量,包括过敏史、肾功能不全和ICM种类,其中过敏史是最主要的影响因素。 结论 通过运用Logistic 回归和决策树模型预测不良反应的影响因素并将研究转变为了避免不良反应发生的主动预测,对患者安全用药有参考价值。 -
关键词:
- 碘造影剂 /
- 不良反应 /
- Logistic回归 /
- 决策树模型
Abstract:Objective Logistic regression and decision tree model were used to analyze the influencing factors of adverse reactions of iodine contrast agents. Methods The adverse reaction group consisted of 129 patients who underwent CT examination using ICM and had adverse reactions from January 2019 to August 2023 in Fuwai Yunnan Hospital. The control group consisted of 135 patients who did not have adverse reactions during the same period using ICM. The variables with statistical significance in the univariate analysis were incorporated into multi-factor Logistic regression and decision tree model to extract and analyze the influencing factors causing adverse reactions. Results Logistic regression showed that BMI, allergic history, and renal insufficiency were the influencing factors of ICM (P < 0.05). The prediction model was: Lggit (P) = 0.123×BMI+1.684×allergy history + 2.551×renal insufficiency −3.652. Decision tree analysis screened out three risk variables, including allergy history, renal insufficiency and ICM type, among which allergy history was the most important influencing factor. Conclusion Logistic regression and the decision tree model can predict the influencing factors of adverse reactions and transform the study into an active prediction to avoid adverse reactions, it has reference value for the safe drug use of patients. -
Key words:
- Iodinated contrast medium /
- Adverse reactions /
- Logistic regression /
- Decision tree model
-
表 1 不良反应组与对照组单因素分析结果[n(%)/M(P25,P75)]
Table 1. Results of one-way analysis of variance between adverse reaction group and control group [n(%)/M(P25,P75)]
变量 n 不良反应组(n = 129) 对照组(n = 135) χ2/Z P 性别 男 161 76(58.9) 85(63.0) 0.454 0.500 女 103 53(41.1) 50(37.0) 年龄(岁) 264 53(43,66) 55(46,67) 0.326 0.568 BMI(kg/m2) 259 24.49(22.14,26.57) 23.12(20.05,25.32) 12.107 0.001* 血钾(mmol/L) 261 4.03(3.80,4.26) 4.07(3.85,4.30) 1.493 0.222 肌酐(μmol/L) 260 75.8(61.5,90.4) 73.4(59.8,87.9) 0.956 0.328 用药时间 第一季度 37 16(12.40) 21(15.56)
7.725
0.052第二季度 66 25(19.38) 41(30.37) 第三季度 89 53(41.09) 36(26.67) 第四季度 72 35(27.13) 37(27.41) ICM种类 碘海醇 145 61(47.29) 84(62.22)
16.973
0.001*碘克沙醇 20 13(10.08) 7(5.19) 碘帕醇 45 17(13.18) 28(20.74) 碘普罗胺 54 38(29.46) 16(11.85) 过敏史 有 62 48(37.21) 14(10.37) 26.443 < 0.001* 无 202 81(62.79) 121(89.63) 吸烟史 有 105 51(39.53) 54(40.0) 0.006 0.938 无 159 78(60.47) 81(60.0) 饮酒史 有 60 32(24.81) 28(20.74) 无 204 97(75.19) 107(79.26) 高血压 有 138 75(58.14) 63(46.67) 3.480 0.062 无 126 54(41.86) 72(53.33) 糖尿病 有 43 16(12.40) 27(20.0) 无 221 113(87.60) 108(80.0) 高血脂 有 129 65(50.39) 64(47.41) 0.234 0.628 无 135 64(49.61) 71(52.59) 冠脉疾病 无 132 69(53.49) 63(46.67) 1.228 0.268 有 132 60(46.51) 72(53.33) 高尿酸血症 有 73 36(27.91) 37(27.41) 0.008 0.928 无 191 93(72.09) 98(72.59) 肝功能不全 有 2 2(1.55) 0(0) 2.109 0.146 无 262 127(98.45) 135(100) 肾功能不全 有 12 11(8.53) 1(0.74) 9.218 0.002* 无 252 118(91.47) 134(99.26) 合用阿司匹林 有 101 43(33.33) 58(42.96) 2.590 0.108 无 163 86(66.67) 77(57.04) 合用α受体阻断剂 有 4 4(3.10) 0(0) 4.250 0.039* 无 260 125(96.90) 135(100) 合用β受体阻滞剂 有 95 41(31.78) 54(40) 1.934 0.164 无 169 88(68.22) 81(60) 合用ACEI/ARB 有 76 33(25.58) 43(31.85) 1.265 0.261 无 188 96(74.42) 92(68.15) *P < 0.05。 表 2 ICM不良反应的变量赋值
Table 2. Variable assignment of ICM adverse reactions
自变量 赋值方式 ICM种类 碘海醇 = 1,碘克沙醇 = 2,碘帕醇 = 3,碘普罗胺 = 4 过敏史 否 = 0,是 = 1 肾功能不全 否 = 0,是 = 1 表 3 多因素Logistic回归分析
Table 3. Multivariate Logistic regression analysis
变量 β S.E. Waldχ2 P OR 95%CI BMI 0.123 0.040 9.595 0.002* 1.131 1.046~1.223 ICM种类 0.107 0.121 0.782 0.377 1.113 0.878~1.412 过敏史 1.684 0.354 22.575 < 0.001* 5.387 2.689~10.791 肾功能不全 2.551 1.081 5.571 0.018* 12.822 1.541~106.664 常量 −3.652 0.897 16.570 < 0.001* 0.026 − *P < 0.05。 -
[1] Böhm I,Morelli J,Nairz K,et al. Myths and misconceptions concerning contrast media induced anaphylaxis: A narrative review[J]. Postgrad Med,2017,129(2):259-266. doi: 10.1080/00325481.2017.1282296 [2] Vega F,Maria Victoria Múgica,Bazire R,et al. Adverse reactions to iodinated contrast media: Safety of a study protocol that includes fast full dose parenteral challenge tests searching for an alternative contrast media[J]. Clin Exp Allergy,2020,50(2):271-274. doi: 10.1111/cea.13518 [3] Park H K,Kang M G,Yang M S,et al. Epidemiology of druginduced anaphylaxis in a tertiary hospital in Korea[J]. Allergol Int,2017,66(4):557-562. doi: 10.1016/j.alit.2017.02.008 [4] Bae K,Lee S M,Ha J Y,et al. Adverse drug reactions to CT contrast media in south Korea: Incidence and risk factors[J]. J Korean Soc Radiol,2016,75(1):41-48. doi: 10.3348/jksr.2016.75.1.41 [5] Li X,Chen J,Zhang L,et al. Clinical observation of the adverse drug reactions caused by non - ionic iodinated contrast media: Results from 109 255 cases who underwent enhanced CT examination in Chongqing,China[J]. Br J Radiol,2015,88(1047):20140491. doi: 10.1259/bjr.20140491 [6] Zhang B C,Hou L,Lv B,et al. Post - marketing surveil-lance study with iodixanol in 20,185 Chinese patients from routine clinical practices[J]. Br J Radiol,2014,87(1034):20130325. doi: 10.1259/bjr.20130325 [7] Suh Y J,Yoon S H,Hong H,et al. Acute adverse reactions to nonionic iodinated contrast media: A meta-analysis[J]. Invest Radiol,2019,54(9):589-599. doi: 10.1097/RLI.0000000000000568 [8] Kaito D,Homma K,Sasaki J. Rapid collapse of the inferior vena cava in a patient with cardiac arrest induced by anaphylactic shock after iodinated contrast medium injection[J]. J Am Coll Emerg Physicians Open,2020,1(4):379-382. doi: 10.1002/emp2.12166 [9] 中华医学会放射学分会对比剂安全使用工作组. 碘造影剂使用指南(第2版)[J]. 中华医学杂志,2014,94(43):3363-3369. doi: 10.3760/cma.j.issn.0376-2491.2014.43.003 [10] 邱海燕,王羚入,李建洁,等. 332683例CT增强检查患者非离子碘对比剂急性不良反应情况及其影响因素分析[J]. 陆军军医大学学报,2023,45(3):257-264. [11] 林小玲,洪张翔,司徒启娇,等. 碘对比剂在肿瘤专科医院CT增强检查中急性不良反应预测模型的构建与评估[J]. 现代临床护理,2022,21(9):13-18. doi: 10.3969/j.issn.1671-8283.2022.09.003 [12] 杨秋月,汪晖,李琴. 非离子型碘对比剂致急性不良反应现状及影响因素的研究进展[J]. 循证护理,2022,8(7):920-924. doi: 10.12102/j.issn.2095-8668.2022.07.012 [13] McDonald J S,Larson N B,Schmitz J J,et al. Acute adverse events after iodinated contrast agent administration of 359,977 injections: A single-center retrospective Study[J]. Mayo Clin Proc,2023,98(12):1820-1830. doi: 10.1016/j.mayocp.2023.02.032 [14] Feng Y,Wang J,Shao Z,et al. Predicting related factors of immunological response to hepatitis B vaccine in hemodialysis patients based on integration of decision tree classification and logistic regression[J]. Hum Vaccin Immunother,2021,17(9):3214-3220. doi: 10.1080/21645515.2021.1895603 [15] 李良俊,翟荣,邬闻文,等. Logistic回归及决策树模型在CCU老年病人睡眠障碍影响因素分析中的应用[J]. 护理研究,2022,36(16):2874-2879. doi: 10.12102/j.issn.1009-6493.2022.16.008 [16] Voltolini S,Cofini V,Murzilli F,et al. Hypersensitivity reactions to iodinated contrast mediain Italy: A retrospective study. Characteristics of patients and risk factors[J]. Eur Ann Allergy Clin Immunol,2022,54(2):60-67. doi: 10.23822/EurAnnACI.1764-1489.225 [17] Iordache A,Docea A,Buga A,et al. The incidence of skin lesions in contrast media induced chemical hypersensitivity[J]. Experimental and Therapeutic Medicine,2019,17(2):1113-1124. [18] Goksel O,Aydin O,Atasoy C,et al. Hypersensitivity reactions to contrast metia: Prevalence,risk factors and the role of skin tests in diagnosis across sectional survey[J]. International Archives of Allergy and Immunology,2011,155(3):297-305. doi: 10.1159/000320760 [19] Abe S,Fukuda H,Tobe K,et al. Protective effect against repeat adverse reactions to iodinated contrast medium: Premedication vs. changing the contrast medium[J]. Euopean Radiology,2016,26(7):2148-2154. [20] Zhai L Q,Guo X J,Zhang H Y,et al. Non-ioniciodinated contrast media related immediate reactions: Amechanism study of 27 patients[J]. Leg Med (Tokyo),2017,24:56-62. doi: 10.1016/j.legalmed.2016.11.006 [21] 胡立禄,高全清,刘振良. 苯海拉明联合泼尼松预防碘造影剂过敏反应的临床观察[J]. 中国药房,2016,27(6):765-767. doi: 10.6039/j.issn.1001-0408.2016.06.16 [22] 姚英. 2018年欧洲泌尿生殖放射学会造影后急性肾损伤防治指南的解读[J]. 中国血液净化,2019,18(6):435-438. doi: 10.3969/j.issn.1671-4091.2019.06.017 [23] An J,Jung H,Kwon O Y,et al. Differences in adverse reactions among iodinated contrast media: Analysis of the KAERS database[J]. J Allergy Clin Immunol Pract,2019,7(7):2205-2211. doi: 10.1016/j.jaip.2019.02.035 [24] 冉超. 碘造影剂不良反应信号挖掘及危险因素分析[D]. 重庆: 重庆医科大学, 2020.