基于多因素logistic回归模型和决策树模型的肺结核患者就诊延迟及确诊延迟影响因素分析

Analysis on influencing factors of medical care seeking delay and diagnosis delay of pulmonary tuberculosis patients based on logistic regression model and decision tree model

目的了解北京市通州区及昌平区肺结核患者就诊延迟及确诊延迟情况并分析其影响因素,为早期发现及科学管理肺结核患者提供有效建议。方法采用回顾性流行病学调查方法,利用全国结核病管理信息系统收集北京市通州区及昌平区于2021年1月1日至12月31日登记的肺结核患者个案资料,采用电话访谈对部分信息进行补充,采用多因素logistic回归模型及决策树模型分析肺结核患者就诊延迟及确诊延迟的影响因素。结果2021年北京市通州区及昌平区肺结核患者就诊延迟时间M(Q1,Q3)为11(5,26)d,就诊延迟率为41.71%。多因素分析结果显示,健康体检(OR=0.033,95%CI:0.008~0.147)、就诊前咳嗽咳痰<2周,且有任一结核病症状(OR=0.378,95%CI:0.215~0.665)、就诊前有其他症状(OR=2.791,95%CI:1.710~4.555)及就诊期间无需上班或上学(OR=2.990,95%CI:1.419~6.298)为就诊延迟的影响因素;肺结核患者确诊延迟时间M(Q1,Q3)为8(0,18)d,确诊延迟率为35.20%。多因素分析结果显示,确诊单位为结核病专科医院(OR=0.426,95%CI:0.236~0.767)和结核病防治所(OR=1.843,95%CI:1.061~3.202)以及患者来源为追踪(OR=2.632,95%CI:1.062~6.521)为确诊延迟的影响因素。多因素logistic回归模型与决策树模型的整体效能相当,决策树模型的灵敏度高于多因素logistic回归模型、特异度低于多因素logistic回归模型。结论2021年北京市通州区及昌平区肺结核患者就诊延迟及确诊延迟处于较低水平,但仍需加强宣传教育并积极开展主动筛查,提高市民对结核病的防治意识,同时进一步提升医疗服务水平和改善就医可及性,减少患者就医及诊断延迟现象。

Objective To investigate the status of medical care seeking delay and diagnosis delay of pulmonary tuberculosis(PTB)patients in Tongzhou District and Changping District of Beijing,analyze the related factors and put forward suggestions for early detection and scientific management of PTB patients.Methods A retrospective epidemiological survey was conducted to collect the incidence data of PTB registered in Tongzhou and Changping from January 1 to December 31,2021 by using the Chinese Tuberculosis Information Management System,and telephone interview were used for information supplement.Multivariate logistic regression model and decision tree model were used to analyze the influencing factors of medical care seeking delay and diagnosis delay of PTB patients.Results In 2021,the medical care seeking delay time M(Q1,Q3)in the PTB patients in Tongzhou and Changping was 11(5,26)days,with a delay rate of 41.71%.Results from multivariate logistic regression model analysis revealed that factors influencing the medical care seeking delay included regular health check-up(OR=0.033,95%CI:0.008-0.147),coughing for less than 2 weeks or showing any symptom of PTB before medical care seeking(OR=0.378,95%CI:0.215-0.665),showing other symptoms before medical care seeking(OR=2.791,95%CI:1.710-4.555),no work or school in medical care seeking(OR=2.990,95%CI:1.419-6.298).The diagnosis delay time M(Q1,Q3)in the PTB patients was 8(0,18)days,with a delay rate of 35.20%.Multivariate logistic regression model analysis revealed that the factors influencing the diagnosis delay of PTB included being diagnosed at a specialized tuberculosis hospital(OR=0.426,95%CI:0.236-0.767)or a tuberculosis prevention and control institution(OR=1.843,95%CI:1.061-3.202)and being traced as a source of infection(OR=2.632,95%CI:1.062-6.521).The overall performance of the multivariate logistic regression model was comparable to that of the decision tree model,with the decision tree model exhibiting higher sensitivity but lower specificity.Conclusions The medical care seeking delay rate and diagnosis delay rate of tuberculosis in Tongzhou and Changping were at low levels in 2021.However,it is still necessary to strengthen the health education and active screening,improve the public awareness of PTB prevention and control,and further improve the level of medical services and medical access to reduce the medical care seeking delay and diagnosis delay of PTB patients.