| 吴晓文,黄银梅,邱潮锋.基于 logistic回归分析探究非小细胞肺癌骨髓抑制的危险因素及预测方法[J].安徽医药,2021,25(5):894-898. |
| 基于 logistic回归分析探究非小细胞肺癌骨髓抑制的危险因素及预测方法 |
| Exploration of risk factors and prediction methods of myelosuppression in non-small cell lung cancer based on logistic regression analysis |
| |
| DOI：10.3969/j.issn.1009-6469.2021.05.012 |
| 中文关键词: 癌，非小细胞肺 logistic回归分析 骨髓抑制 危险因素 预测模型 |
| 英文关键词: Carcinoma, non-small-cell lung Logistic regression analysis Myelosuppression Risk factors Prediction model |
| 基金项目: |
|
| 摘要点击次数: 2212 |
| 全文下载次数: 774 |
| 中文摘要: |
| 目的基于 logistic回归分析探讨非小细胞肺癌( NSCLC)化疗发生骨髓抑制的危险因素并建立预测模型。方法以 2017年 9月至 2019年 12月汕头市潮阳区大峰医院收治的 120例 NSCLC病人为研究对象，以Ⅱ级及以上骨髓抑制的病人定为观察组( 52例)I级及以下的病人定为对照组(68例)对两组病人的一般资料和治疗情况进行单因素分析，对两组差异有统计学意义的单因素进，行非条件 logistic多因素回归分析，，探究 NSCLC病人发生骨髓抑制的危险因素并建立预测模型。重新纳入 100例 NSCLC病人采用该模型进行预测，计算该模型的敏感度和特异度，并采用 ROC曲线判定其预测价值。结果两组在年龄、体质量指数( BMI)、肿瘤临床分期、是否发生骨髓 /骨转移、治疗前白细胞是否降低( <4.0×109)、轻度白蛋白水平低下( 30~40 g/L)、曾接受化疗次数、 3周内是否进行了放疗上差异有统计学意义( P<0.05)，在性别、吸烟史、化疗方案上差异无统计学意义(P>0.05)。经 logistic回归分析，年龄、发生骨髓 /骨转移、治疗前白细胞降低、轻度白蛋白水平低下、化疗方案等 5个指标 /变量为骨髓抑制的影响因素( P<0.05)。 logistic回归模型: logit(P)= 0.486+1.213X1+1.781X2 +0.611X3+0.682X4-0.246X5，其中自变量: X1为年龄( ≥60岁)X2为发生骨髓 /骨转移， X3为治疗前白细胞数 <4.0×109X4为 TP化疗方案， X5为 GP或 AP化疗方案。基于该模型进行 ROC分析，，其 AUC为 0.812，标准误 0.028，95%CI:0.758~0.867。对，重新纳入的 100例 NSCLC病人进行分析，其预测的敏感度为 92.68%，特异度 83.05%，Kappa系数为 0.806。结论年龄、发生骨髓 /骨转移、治疗前白细胞降低、白蛋白水平低下、 3周内进行了放疗是骨髓抑制的危险因素，该模型有望用于化疗前风险预测，对高危病人进行干预治疗。 |
| 英文摘要: |
| Objective To study the risk factors and prediction methods of myelosuppression in non-small cell lung cancer (NSCLC) based on logistic regression analysis.Methods A total of 120 NSCLC patients admitted to Dafeng Hospital, Chaoyang District, Shantou City from September 2017 to December 2019 were selected as study objectives. Patients with myelosuppression of grade Ⅱ andabove were selected as the observation group (52 cases), and those with I and below were designated as control group (68 cases). The basic information and treatment of two groups were compared for single factor analysis, and the significantly different single factor wereanalyzed by non-conditional logistic regression analysis, the risk factors for myelosuppression and the forecast model were established.A total of 100 new NSCLC patients were forecasted by this model. The sensitivity and specificity of the model were calculated, and thepredictive value was determined by ROC curve.Results There were statistically significant differences in age, BMI, clinical stage oftumor, bone marrow / bone metastasis, leukocyte reduction before treatment (< 4.0 × 109), low level of albumin (30-40 g/ L), times of chemotherapy, whether radiotherapy was carried out within 3 weeks (P < 0.05). There were no statistically significant differences in gender, smoking history and chemotherapy plan (P > 0.05). According to the logistic regression analysis, age, bone marrow / bone metastasis, leucopenia before treatment, low level of albumin and chemotherapy were the influencing factors (P < 0.05). Logistic regression model: logit (P) = 0.486 + 1.213x1 + 1.781x2 + 0.611x3 + 0.682x4-0.246x5, of which independent variable: X1 was age (≥ 60 yearsold), X2 was bone marrow / bone metastasis, X3 was leucocyte number before treatment < 4.0 × 109, X4 was TP chemotherapy, X5 wasGP or AP chemotherapy. ROC analysis based on this model showed that AUC was 0.812, standard error is 0.028, 95% CI: 0.758 ~0.867. The sensitivity, specificity and kappa coefficient were 92.68%, 83.05% and 0.806, respectively.Conclusion Age, bone marrow/ bone metastasis, leukocyte reduction before treatment, low albumin level, and radiotherapy within 3 weeks are risk factors for myelosuppression. This model is expected to be used for pre-chemotherapy risk prediction and intervention therapy for high-risk patients. |
|
查看全文
查看/发表评论 下载PDF阅读器 |
| 关闭 |
|
|
|
