| 张盼,魏媛媛,王浩,等.糖尿病足感染病人多药耐药的危险因素分析及列线图预测模型的建立与验证[J].安徽医药,2026,30(3):556-561. |
| 糖尿病足感染病人多药耐药的危险因素分析及列线图预测模型的建立与验证 |
| Risk factors for multidrug resistance in patients with diabetic foot infection and establishment and validation of a nomogram prediction model |
| |
| DOI：10.3969/j.issn.1009-6469.2026.03.025 |
| 中文关键词: 糖尿病足 伤口感染 多药耐药 LASSO-logistic回归分析 危险因素 列线图 |
| 英文关键词: Diabetic foot Wound infection Multidrug resistance LASSO-logistic regression analysis Risk factors Nomo. gram |
| 基金项目:河北省医学科学研究课题计划( 20210852) |
|
| 摘要点击次数: 306 |
| 全文下载次数: 192 |
| 中文摘要: |
| 目的分析糖尿病足感染病人多药耐药的危险因素，建立并验证列线图预测模型。方法选取 2021年 1月至 2023年 12月于石家庄市人民医院进行治疗的糖尿病足感染病人 113例作为建模组(根据药敏试验结果是否多药耐药分为多药耐药组 42例与对照组 71例)以 2020年 1—12月治疗的糖尿病足感染病人 100例作为验证组。收集病人临床资料，在建模组采用单因素及多因素 logistic回，归分析探索糖尿病足感染病人多药耐药的影响因素，并建立列线图风险预测模型，先对其进行内部验证，再引用验证组数据对其进行外部验证。结果建模组 42例多药耐药病人共检测出多药耐药菌 71株，革兰阴性菌［大肠埃希菌 15.49%(11/71)、铜绿假单胞菌 22.54%(16/71)为主］与革兰阳性菌［金黄色葡萄球菌 19.72%(14/71)、表皮葡萄球菌 7.04%(5/71)、粪肠球菌 8.45%(6/71)为主］分别占比 57.75%(41/71)、 42.25%(30/71)。相较于对照组，多药耐药组病人骨髓炎［61.90%(26/42)比 42.25%(30/71)］、神经缺血性伤口［66.67%(28/42)比 46.48%(33/71)］、长期使用抗菌药物［95.24%(40/42)比 46.48%(33/71)］、抗菌药物联合使用种数 >2种［78.57%(33/42)比 38.03%(27/71)］占比明显增加，超敏 C反应蛋白( hs-CRP)水平［(22.87±4.57)mg/L比(9.35±1.26)mg/L］明显升高( P<0.05)。基于 LASSO-logistic回归分析，骨髓炎、神经缺血性伤口、长期使用抗菌药物、抗菌药物联合使用种数、 hs-CRP均为建模组糖尿病足感染病人多药耐药的独立影响因素(P<0.05)。建模组受试者操作特征曲线(ROC曲线)下面积(AUC)为 0.82［95%CI:(0.79，0.84)］约登指数最大为 0.63，此时对应的模型预测概率为 0.380，灵敏度为 85.30%，特异度为 78.00%，Hosmer-Lemeshow检验 P=0.215；验证，组 AUC为 0.83［95%CI:(0.79，0.90)］，Hosmer-Lemeshow检验 P=0.162，列线图风险预测模型对糖尿病足感染病人多药耐药具有较高的预测能力。结论基于骨髓炎、神经缺血性伤口、长期使用抗菌药物、抗菌药物联合使用种数、 hs-CRP等糖尿病足感染病人多药耐药的独立影响因素建立列线图风险预测模型，准确度较高，且具有较高的预测价值。 |
| 英文摘要: |
| Objective To analyze the risk factors for multidrug resistance (MDR) in patients with diabetic foot infection (DFI) and toestablish and validate a nomogram prediction model.Methods A total of 113 DFI patients treated at Shijiazhuang People's Hospitalfrom January 2021 to December 2023 were selected as the modeling cohort [divided into an MDR group (n=42) and a control group (n= 71) based on antimicrobial susceptibility testing results]. Another 100 DFI patients treated from January to December 2020 served asthe validation cohort. Clinical data were collected. Univariate and multivariate logistic regression analyses were used in the modelingcohort to explore factors influencing MDR in DFI patients. A nomogram prediction model was constructed, underwent internal valida.tion, and was subsequently externally validated using the validation cohort data. Results In the modeling cohort, 42 MDR patients yielded 71 MDR bacterial strains. Gram-negative bacteria [primarily Escherichia coli 15.49% (11/71) and Pseudomonas aeruginosa 22.54% (16/71)] and Gram-positive bacteria [primarily Staphylococcus aureus 19.72% (14/71), Staphylococcus epidermidis 7.04% (5/ 71), and Enterococcus faecalis 8.45% (6/71)] accounted for 57.75% (41/71) and 42.25% (30/71), respectively. Compared to the controlgroup, the MDR group had significantly higher proportions of patients with osteomyelitis [61.90% (26/42) vs. 42.25% (30/71)], neu. roischemic wounds [66.67% (28/42) vs. 46.48% (33/71)], long-term antibiotic use [95.24% (40/42) vs. 46.48% (33/71)], and combina. tion use of >2 types of antibiotics [78.57% (33/42) vs. 38.03% (27/71)], along with significantly elevated high-sensitivity C-reactive pro. tein (hs-CRP) levels [(22.87±4.57) mg/L vs. (9.35±1.26) mg/L] (all P<0.05). Based on LASSO-logistic regression analysis, osteomyelitis, neuroischemic wound, long-term antibiotic use, number of combined antibiotics used, and hs-CRP were identified as independent influ. encing factors for MDR in the modeling cohort (all P<0.05). The area under the ROC curve (AUC) for the modeling cohort was 0.82 [95% CI: (0.79, 0.84)]. The maximum Youden index was 0.63, corresponding to a model prediction probability cutoff of 0.380, with asensitivity of 85.30% and specificity of 78.00%. The Hosmer-Lemeshow test yielded P=0.215. For the validation set, the AUC was 0.83 [95% CI: (0.79, 0.90)], and the Hosmer-Lemeshow test yielded P=0.162, indicating that the nomogram prediction model possesses a high predictive ability for MDR in DFI patients.Conclusion The nomogram prediction model, established based on independent influ.encing factors for MDR in DFI patients including osteomyelitis, neuroischemic wound, long-term antibiotic use, number of combined antibiotics used, and hs-CRP, demonstrates high accuracy and possesses substantial predictive value. |
|
查看全文
查看/发表评论 下载PDF阅读器 |
| 关闭 |
|
|
|
