基层集中医学观察点医护人员疲劳影响因素分析及预测
Analysis and prediction of influencing factors of fatigue of medical staff in primary centralized medical observation points
曲英娜1,郑曙光1,杨婷2,郑丽妮2,杨飞2,江芬2
Qu Yingna1, Zheng Shuguang1, Yang Ting2, Zheng Lini2, Yang Fei2, Jiang Fen2
1浙江省金华市兰溪市人民医院科研教学部 321100;2浙江省金华市兰溪市人民医院321100;
1Department of Scientific Research and Teaching, Lanxi People's Hospital, Jinhua City, Zhejiang Province, 321100, China; 2. Lanxi People's Hospital, Jinhua City, Zhejiang Province 321100;
【摘 要】目的 探讨集中医学观察点医护人员疲劳的影响因素并建立预测模型,为疲劳的预测提供参考,并提供减少医护人员疲劳的可行性建议。方法 选取2021年1月至2023年4月兰溪市人民医院参与基层疫情防控145医护工作者研究对象,以问卷形式收集医护人员个人及工作信息。依照观察对象的MFI-20得分,分析疲劳组(≥60分)与对照组(<60分)间不同变量的差异是否有统计学意义。基于差异显著变量,建立logistic回归模型,绘制预测疲劳状态的ROC曲线,并与单一显著变量预测疲劳状态的ROC曲线进行对比分析,检验模型预测效能。结果 二元logistic多因素回归分析结果显示,夜班占工作总时长占比、是否参加体育锻炼、性别均为影响疲劳的独立影响因素(P<0.05)。应用二元logistic回归模型的建立统计学诊断模型(AUC=0.739,敏感度=66.0%)优于单独应用夜班占比的ROC分析(AUC=0.605,敏感度=48.9%),诊断价值较高。结论 夜班占工作总时长占比、是否参加体育锻炼、性别均影响疲劳的发生,其联合诊断模型预测值(Pre)有利于疲劳诊断,从而帮助医院制定相应预防对策。
【Abstract】ObjectiveTo explore the influencing factors of fatigue among medical staff in centralized medical observation points and to establish a prediction model, so as to provide a reference for the prediction of fatigue and to reduce the fatigue of medical staffFeasibility proposals. Methods From January 2021 to April 2023, 14 5 medical workers in Lanxi People's Hospital who participated in the prevention and control of the epidemic at the grassroots level were selected, and the personal and work information of medical staff were collected in the form of questionnaires. According to the MFI-20 score of the observed subjects, the differences of different variables between the fatigue group (≥60 points) and the control group (<60 points) were analyzed to see if there was any statistical significance. Based on the differentially significant variables, a logistic regression model was established, and the ROC curve for predicting fatigue state was plotted, and the ROC curve for predicting fatigue state was compared with that of a single significant variable to test the prediction performance of the model. Results The results of binary logistic multivariate regression analysis showed that the proportion of night shift in total working hours, whether or not to participate in physical exercise, and gender were independent influencing factors for fatigue (P<0.05). )。 The establishment of a statistical diagnostic model using the binary logistic regression model (AUC=0.739, sensitivity=66.0%) was better than the ROC analysis of the proportion of night shifts alone (AUC= 0.605, sensitivity=48.9%), and the diagnostic value was high. Conclusion The proportion of night shift in the total working hours, whether or not to participate in physical exercise, and gender all affect the occurrence of fatigue, and the predictive value (Pre) of the joint diagnostic model is conducive to the diagnosis of fatigue, so as to help hospitals formulate corresponding preventive measures.
【关键词】疲劳;影响因素研究;ROC曲线分析
【Keywords】fatigue; influencing factors research; ROC curve analysis
Analysis of Influencing Factors and Prediction of Fatigue among Medical Staff at Primary Centralized Medical Observation Points
QU Ying-na,ZHENG Shu-guang,YANG Ting,ZHENG Li-ni,
JIANG Fen,YANG Fei
Lanxi People’s Hospital Lanxi, Jinhua, Zhejiang, 321100,China
[Abstract] Objective To explore the influencing factors of fatigue of medical staff in large centralized medical observation points , so as to provide a diagnostic scheme for fatigue prediction and feasible suggestions for reducing fatigue . Methods 145 medical staff from Lanxi People’s Hospital who participated in grassroots epidemic prevention and control from January 2021 to April 2023 were selected. Basic personal and job information were collected by questionnaire. Review the Multi-Dimensional Fatigue Scale (MFI-20) score and other information after the closed loop cycle (at least 14+7 days). Whether the total score of MFI-20 of the observation subjects was divided into the obvious fatigue group (≥60 points) and the control group (<60 points), and the differences variables between the two groups were analyzed to see whether there was statistical significance. A logistic regression model was established to draw the ROC curve for predicting fatigue state with different variables, and the ROC curve for predicting fatigue state with a single significant variable was compared and analyzed to test the prediction efficiency of the model. Results The results of binary logistic regression analysis showed that the proportion of night shift in total working hours, whether to participate in physical exercise and gender were all independent influencing factors on fatigue (P<0.05). The predictive value of statistical diagnosis model (Pre) was established by Logistic regression model combined with meaningful variables, which was superior(AUC=0.739, sensitivity =66.0%) to ROC curve analysis (AUC=0.605, sensitivity =48.9%) when night shift proportion was applied alone to predict fatigue which means the diagnostic value was high. Conclusion The proportion of night shift in total working hours, whether to participate in physical exercise and gender all affect the occurrence of fatigue, and the combined diagnostic model predictive value (Pre) is beneficial to the diagnosis of fatigue, so as to help hospitals develop corresponding preventive measures.
Key Words :Fatigue;Study on influencing factors;ROC curve analysis;
集中医学观察点拥有单体面积大、观察人员容量多、工作环境半露天、单间自然通风效果好等有利条件。根据浙江省疫情防控管理规定,集中医学观察点医护人员在工作期间全部实行闭环管理[1],医护人员在污染区上班时需全程穿防护服。长时间的闭环管理与高强度工作,使医护人员的疲劳状况加重[2],成为医护人员精力过度消耗[3]的不利因素。如何在高强度疫情防控工作下,优化隔离点工作设置,减少疲劳感,进而助力县级基层医疗机构高效完成基层首诊的重要职责[4],成为打赢疫情防控战役的重要举措。当前集中医学观察点疲劳相关研究主要关注隔离病房医护人员的精神压力情况[5]和心理方面[6]研究,忽视了工作环境、时长、流程等现实可控因素对医护人员疲劳的影响。因此本研究基于大型方舱式集中医学观察点,调查基层医疗医护人员在闭环管理期间的疲劳状态与健康状况。使用二元logistic回归模型[7]探讨观察点医护人员疲劳的影响因素并建立预测模型[8],提供预测和降低疲劳程度的富有建设性举措。
The centralized medical observation point has favorable conditions such as large single area, large capacity of observers, semi-open working environment, and good natural ventilation effect of single rooms. According to the Zhejiang Provincial Regulations on Epidemic Prevention and Control, all medical staff in centralized medical observation sites are subject to closed-loop management during their work[1], and medical staff are required to wear protective clothing throughout their work in contaminated areas. Long-term closed-loop management and high-intensity work aggravate the fatigue of medical staff [2], and they become medical staffDisadvantage of excessive energy consumption [3]. How to optimize the work setting of isolation points and reduce fatigue under the high-intensity epidemic prevention and control work, so as to help county-level primary medical institutions efficiently complete the important responsibilities of primary diagnosis at the grassroots level [4].It has become an important measure to win the battle of epidemic prevention and control. At present, research on fatigue in centralized medical observation points mainly focuses on the mental stress and psychological aspects of medical staff in isolation wards [5].[6] The study ignores the impact of realistic and controllable factors such as working environment, duration, and process on the fatigue of medical staff. Therefore, this study investigated the fatigue state and health status of primary medical staff during closed-loop management based on a large-scale centralized medical observation point in the form of a shelter. A binary logistic regression model [7] was used to explore the influencing factors of fatigue among healthcare workers at observation sites and a prediction model was established [8]. ] to provide constructive measures to predict and reduce fatigue levels.
1对象与方法
1. Objects and Methods
1.1研究对象
1.1 Research Subjects
选取2021年1月至2023年4月兰溪市人民医院参与疫情防控145工作者研究对象。研究对象入选标准为:1.基层医护人员;2.从事大型方舱式集中医学观察点工作;3.完成一个闭环管理周期(≥14+7 d)。排除标准为:1.有躯体疾病或进入闭环管理前已经处于疲劳状态;2.正在服用精神类药物。
From January 2021 to April 2023, 14 5 workers from Lanxi People's Hospital who participated in epidemic prevention and control were selected. The inclusion criteria for the study subjects were: 1primary care workers; 2.Engaged in the work of large-scale shelter-type centralized medical observation points; 3.Complete a closed-loop management cycle (≥14+7 d). The exclusion criteria are:1Have a physical illness or are already in a state of fatigue before entering closed-loop management; 2.Are taking psychotropic medications.
1.2调查工具
1.2 Survey Tools
问卷主要收集参与疫情防控医护人员1.基本情况:包括性别、年龄、学历、婚姻状况、职称、是否独生子女、执业类别;2.工作环境情况:单位性质、进入隔离点工作次数、隔离点性质、夜班占工作总时长占比;3.运动情况:是否规律参加体育锻炼;4.疾病情况:闭环周期结束后(至少14+7d)MOSDS问卷得分获取。
The questionnaire mainly collects the basic information of medical staff involved in epidemic prevention and control: including gender, age, education, marital status, professional title, whether they are only children, and practice category; 2. Working environment: the nature of the unit, the number of times you enter the isolation point, the nature of the isolation point, and the proportion of night shift in the total working hours; 3. Exercise: whether to participate in physical exercise regularly; 4. Disease status: After the end of the closed-loop cycle (at least 14+7d), the MOSDS questionnaire score was obtained.
器官对应的症状或疾病状态拟通过自编的多器官症状或疾病评分量表MOSDS[9](multiple organ symptom or disease score):由具备高级职称的医疗专家组建专家组,经集体商议决定。对前期报告的一些症状与疾病进行有针对性的诊断,制成有可操作性的评分量表。主要包括呼吸、消化、心血管、眼耳鼻喉、皮肤、神经精神、泌尿、口腔8个部分。有其中1类对应的症状但不影响工作或睡眠赋1分,有1类症状同时已经影响工作或睡眠赋2分。最低分0分,最高分16分。
The symptoms or disease states corresponding to the organs are proposed to be established by a self-developed multi-organ symptom or disease score scale (MOSDS) [9]: established by a medical expert group with a senior professional titleThe expert group shall decide after collective deliberation. Targeted diagnosis of some symptoms and diseases reported in the early stage was carried out, and a workable rating scale was made. It mainly includes 8 parts: respiratory, digestive, cardiovascular, eye, ear, nose and throat, skin, neuropsychiatric, urinary and oral cavity. 1 point is assigned to having symptoms corresponding to one of these categories but not affecting work or sleep, and 2 points are assigned to having 1 type of symptom that has also affected work or sleep. The minimum score is 0 points, and the maximum score is 16 points.
1.3 疲劳评估
1.3 Fatigue assessment
结合当前疲劳体力脑力心理病理分类法[10],采用MFI-20问卷[11]中文版(Chinese version of the Multidimensional Fatigue Inventory-20)综合评估[12]疲劳得分情况。涵盖体力疲劳、脑力疲劳、动力下降、活动减少四个维度,体力性疲劳主要指机体过度运动后产生的酸胀、乏力感;脑力性疲劳主要指长时间脑力高速运转导致的疲劳;心理性疲劳主要指由于环境变化产生的情绪低落;病理性疲劳则是由于疾病继发性引起的倦怠无力感。疲劳程度越严重量表总分越高,总分100分。闭环周期结束后(至少14+7d)回顾多维疲劳量表(MFI-20)评分,并将观察对象的MFI-20评分总分是否大于等于60分,分为两组,明显疲劳组(≥60分)与对照组(小于60分),分析两组间不同变量的差异是否有统计学意义。
Combined with the current psychopathological classification of fatigue, physical strength, and mental strength [10], the MFI-20 questionnaire was used [11].Chinese version of the Multidimensional Fatigue Inventory-20 (Chinese version of the Multidimensional Fatigue Inventory-20) comprehensive assessment[12] fatigue score. It covers four dimensions: physical fatigue, mental fatigue, decreased motivation, and reduced activity, and physical fatigue mainly refers to the feeling of soreness and fatigue produced by excessive exercise. Mental fatigue mainly refers to fatigue caused by long-term high-speed mental work; Psychological fatigue mainly refers to low mood due to environmental changes; Pathological fatigue is a feeling of fatigue and weakness caused by secondary illness. The more severe the fatigue, the higher the total score of the scale, with a total score of 100. After the end of the closed-loop cycle (at least 14+7 days), the Multidimensional Fatigue Scale (MFI-20) score was reviewed, and whether the total score of the MFI-20 score of the observed subjects was greater than or equal to 60 points, divided into two groups, the obvious fatigue group (≥60 points) and the control group (less than 60 points), and whether the differences between the two groups were statistically significant.
1.4统计学方法
1.4 Statistical Methods
采用SPSS 26.0统计学软件进行数据处理。计数资料以例数或百分率表示,组间比较采用χ2 检验;计量资料均不符合正态分布,采用中位数(四分位数间距)“ M(Q1,Q3)”表示。对于无序分类变量,采用随机二分类卡方检验,对于有序分类变量,多组间有序变量比较采用 Kruskal-Wallis H 检验。对于计量数据采取非参数秩和检验。采用多因素Logistic回归分析探讨疲劳影响因素;联合有意义项目(P<0.05),建立统计学诊断模型,计算预测值(Pre),以Pre值进行ROC曲线分析(ROC curve analysis),并通过受试者工作特征曲线下面积(Area under curve, AUC)评估模型的区分度,验证预测模型的准确性,P<0.05为差异具有统计学意义。
SPSS 26.0 statistical software was used for data processing. Count data were expressed as cases or percentages, and the chi-� 2 test was used for comparison between groups. None of the measures were normally distributed and were expressed as median (interquartile range) "M(Q1, Q3)". For disordered categorical variables, a random dichotomous chi-square test was used, and for ordinal categorical variables, the Kruskal-Wallis H test was used for comparison of ordinal variables between multiple groups . A non-parametric rank-sum test is used for the measured data. Multivariate logistic regression analysis was used to explore the influencing factors of fatigue. Combined with the meaningful project (P<0.05), a statistical diagnostic model was established, the predicted value (Pre) was calculated, the ROC curve analysis was performed with the Pre value, and the area under the receiver operating characteristic curve (Area under curve, AUC) to evaluate the discrimination of the model and verify the accuracy of the prediction model, P<0.05 was statistically significant.
2结果
2 results
2.1数据统计及两组数据比较
2.1 Data statistics and comparison of two sets of data
本次研究共调查145名医护人员,140医护人员全部完成问卷调查,问卷的有效回收率96.6%。从人员构成看,本研究样本中医生36人占总人数的35.7%,护士74人占52.8%,医技19人,行政11人。医生和护士群体占88.5%,为观察点抗疫主力。从学历结构来看,大专 29人,占比20.7%,本科105人,占比75%,本科学历人数最多。从隔离点性质来看,在酒店或宾馆医护人员81人,占比57.8%,方舱、工厂或学校一共59人,占比42.2%。
In this study, a total of 145 medical staff were surveyed, and all 140 medical staff completed the questionnaire, and the effective recovery rate of the questionnaire was 96.6%. In terms of personnel composition, 36 doctors accounted for 35.7% of the total number of doctors, 74 nurses accounted for 52.8%, 19 medical technicians and 11 administrative personnel. Doctors and nurses accounted for 88.5% of the total, making them the main force in the fight against the epidemic at the observation point. From the perspective of educational structure, there are 29 junior colleges, accounting for 20.7%, and 105 undergraduates, accounting for 75%, with the largest number of bachelor's degrees. From the perspective of the nature of isolation points, there were 81 medical staff in hotels or guesthouses, accounting for 57.8%, and 59 people in shelters, factories or schools, accounting for 42.2%.
依据回顾多维疲劳量表(MFI-20)93例得分60分以下定义为不疲劳,占比66.4% ;47例得分60分(含)以上定义为疲劳,占比33.6%。具体单因素分析结果见表1。非疲劳组男性11名(占比11.8%),女性82(占比88.2%),疲劳组男性17名(占比36.2%),女性30名(63.8%),具有0.01水平(p=0.001)显著差异;非疲劳组夜班占总工作时长比值中位数为27.8%,疲劳组为40%,具有0.05水平(p=0.042)显著差异;非疲劳组参加体育锻炼有27名(占比29.0%),不参加体育锻炼66名(占比71.0%),疲劳组参加体育锻炼22名(占比46.8%),不参加体育锻炼25名(占比53.2%),具有0.05水平(p=0.037)显著差异。
According to the Review Multidimensional Fatigue Scale (MFI-20), 93 patients with a score of less than 60 were defined as no fatigue, accounting for 66.4%; Fatigue was defined in 47 cases with a score of 60 or more, accounting for 33.6%. The results of the specific univariate analysis are shown in Table 1. There were 11 males (11.8%), 82 females (88.2%) in the non-fatigue group, 17 males (36.2%) and 30 females (63.8%) in the fatigue group, with a significant difference of 0.01 level (P=0.001). The median ratio of night shift to total working hours in the non-fatigue group was 27.8% and 40% in the fatigue group, with a significant difference of 0.05 level (P=0.042). There were 27 students (29.0%) in the non-fatigue group, 66 (71.0%) in the non-fatigue group, 22 (46.8%) in the fatigue group, and 25 (53.2%) in the fatigue group, with a significant difference of 0.05 level (P=0.037).
表1 导致疲劳发生可能危险因素的单因素分析 | |||||
因素 | 非疲劳组(n=93) | 疲劳组(n=47) | χ2/Z | P | |
基本情况 | |||||
性别 | 男 | 11(11.8%) | 17(36.2%) | 11.563 | 0.001 |
女 | 82(88.2%) | 30(63.8%) | |||
年龄 | 32(27,41) | 31(26,38) | -0.546 | 0.585 | |
婚姻状况 | 未婚 | 52(55.9%) | 30(63.8%) | 0.806 | 0.369 |
已婚 | 41(44.1%) | 17(36.2%) | |||
学历 | 中专 | 2(2.2%) | 2(4.3%) | 0.005 | 0.944 |
大专 | 20(21.5%) | 9(19.1%) | |||
本科 | 70(75.3%) | 35(74.5%) | |||
硕士研究生 | 1(1.1%) | 1(2.1%) | |||
是否独生子女 | 是 | 22(23.7%) | 14(29.8%) | 0.614 | 0.433 |
否 | 71(76.3%) | 33(70.2%) | |||
工作环境 | |||||
单位性质 | 总院 | 72(77.4%) | 33(70.2%) | 0.865 | 0.352 |
院区 | 21(22.6%) | 14(29.8%) | |||
隔离点性质 | 方舱 | 22(23.7%) | 11(23.4%) | 0.128 | 0.938 |
酒店或宾馆 | 53(57.0%) | 28(59.6%) | |||
工厂或学校 | 18(19.4%) | 8(17%) | |||
职称 | 未定级 | 7 (7.5%) | 7(14.9%) | 0.528 | 0.467 |
初级 | 42 (45.2%) | 19(40.4%) | |||
中级 | 32(34.4%) | 16(34.0%) | |||
高级 | 12 (12.9%) | 5(10.6%) | |||
执业类别 | 医师 | 23(24.7%) | 13(27.7%) | 1.431 | 0.709 |
护士 | 52(55.9%) | 22(46.8%) | |||
医技 | 11(11.8%) | 8(17.0%) | |||
行政 | 7(7.5%) | 4(8.5%) | |||
进入隔离点工作次数 | 1(1,2) | 1(1,2) | -0.082 | 0.934 | |
夜班占工作总时长占比 (%) | 27.8(17.2,39.4) | 40(18.5,23.3) | -2.035 | 0.042 | |
运动情况 | |||||
是否参加体育锻炼 | 是 | 27(29.0%) | 22(46.8%) | 4.337 | 0.037 |
否 | 66(71.0%) | 25(53.2%) | |||
疾病情况 | |||||
MOSDS得分 | 0(0,2) | 0(0,2) | -0.075 | 0.940 |
2.2影响疲劳的多因素
2.2 Multiple factors affecting fatiguelogistic回归分析
regression analysis
以疲劳情况作为因变量(是=1,否=0),以单因素分析中P<0.05的因素(性别、夜班占工作总时长占比、是否参加体育锻炼)作为自变量进行多因素logistic回归分析。其中,性别和是否参加体育锻炼分别设置哑变量。霍斯默-莱梅肖检验的原定假设为:模型拟合值和观测值的吻合程度一致;这里p值大于0.05(χ²=3.031,p=0.932>0.05),因而说明接受原定假设,即说明本次模型通过HL检验,模型拟合优度较好。
Fatigue was used as the dependent variable (yes=1, no=0), and the factors of P<0.05 in the univariate analysis (gender, proportion of night shift in total working hours, and participation in physical exercise) were used as independent variables. Among them, gender and participation in physical exercise were set as dummy variables. The original assumption of the Hosmer-Lemeshaw test was that the fit of the model was consistent with the observed values; Here, the p-value is greater than 0.05 (χ²=3.031, p=0.932>0.05), which indicates that the original hypothesis is accepted, that is, the model has passed the HL test and the model has good goodness of fit.
表2 影响疲劳的多因素logistic回归分析 | |||||||
变量 | β | 标准误差 | 瓦尔德 | P | OR值 | 95% CI | 霍斯默-莱梅肖检验 |
是否规律参加体育锻炼 (0=否 1=是) | 0.993 | 0.420 | 5.593 | 0.018 | 2.699 | 1.185 - 6.146 | χ² =3.031 p=0.932 |
夜班占工作总时长占比 | 2.924 | 0.983 | 8.843 | 0.003 | 18.614 | 2.709 - 127.884 | |
性别(1=男 2=女) | -1.646 | 0.476 | 11.962 | 0.001 | 0.193 | 0.076 - 0.490 |
模型公式为:
The formula of the model is:ln(p/1-p)= -0.761+2.924×夜班占工作总时长占比+0.993
Night shift as a percentage of total working hours +0.993×是否规律参加体育锻炼-1.646
Regular participation in physical activity - 1.646×性别(其中p代表是否疲劳为1的概率,1-p代表是否疲劳为0的概率)。
Gender (where p represents the probability of 1 fatigue and 1-p represents the probability of 0 fatigue).回归结果显示(见表2),
The regression results show (see Table 2),夜班占工作总时长占比
Night shifts as a percentage of the total hours worked、是否参加体育锻炼、性别
Whether or not you participate in physical activity, gender、为影响疲劳
for the effect of fatigue的独立影响因素(P<0.05)。具体而言,
(P<0.05). Specifically,夜班占工作总时长占比的回归系数值为2.924,并且呈现出0.01水平的显著性(z=2.974,p=0.003<0.01),意味着夜班占工作总时长占比会对是否疲劳产生显著的正向影响关系,优势比(OR值)为18.614,意味着夜班占工作总时长占比增加一个单位时,疲劳的增加幅度为18.614倍。是否参加体育锻炼回归系数值为0.993
The regression coefficient value of the proportion of night shift in total working hours was 2.924, and showed a significance of 0.01 level (z=2.974, p=0.003<0.01), which means that the proportion of night shift in total working hours has a significant positive impact on fatigue, and the odds ratio (OR value) is 18.614, which means that when the proportion of night shift in total working hours increases by one unit, the increase in fatigue is 18.614 times. The regression coefficient value of physical exercise was 0.993,呈现0.05水平的显著性(z=0.993,p=0.018<0.05),说明
The significance of the 0.05 level was present (z=0.993, p=0.018<0.05), which was illustrated相对于不参加体育锻炼的人员,参加体育锻炼疲劳的概率是不参加体育锻炼疲劳的2.699倍数;性别的
The probability of fatigue from participating in physical activity was 2.699 times that of fatigue from non-physical activity compared with those who did not participate in physical activity;回归系数值为-1.646,呈现出0.01水平的显著性(z=-1.646,p=0.001<0.01),说明
The regression coefficient value was -1.646, showing a significance of 0.01 (z=-1.646, p=0.001<0.01).相较于男性,女性的疲劳概率是男性的0.193倍。
Women are 0.193 times more likely to be fatigued than men.
2.3模型预测诊断价值
2.3 Model predicts diagnostic value
依据根据单因素分析结果,连续性变量夜班占工作总时长占比可能是疲劳发生的危险因素,纳入并单独进行ROC曲线分析。结果显示,P值为0.042,小于0.05,说明运用夜班占工作总时长预测疲劳可行,曲线下面积AUC为0.605,95%CI为 0.499 ~ 0.712,表明诊断结果具有一定参考价值。依据ROC曲线结果,选取最大正确诊断指数(Youden Index)时的分界值为诊断阈值为40.83%(敏感度为48.9%,特异度为78.5%)。
According to the results of univariate analysis, the proportion of continuous variable night shift in total working hours may be a risk factor for fatigue, and the ROC curve analysis was included and performed separately. The results showed that the P value was 0.042, less than 0.05, indicating that it was feasible to predict fatigue by using night shift as a proportion of total working hours, and the AUC of the area under the curve was 0.605, and the 95%CI was 0.499 ~ 0.712, indicating that the diagnostic results had certain reference value. According to the results of the ROC curve, the cut-off value of the maximum correct diagnostic index (Youden Index) was 40.83% (sensitivity 48.9%, specificity 78.5%).
图1 夜班占比预测疲劳的 ROC 曲线
Figure 1: ROC curve for predicting fatigue as a percentage of night shifts
以Logistic回归模型统计分析数据,计算疲劳预测的概率值(Pre),利用Pre值进行ROC曲线分析。按照诊断概率Logit(P)绘制预测学疲劳的ROC曲线。结果显示,P值为0.000,小于0.01,说明Logit(P)对于疲劳有显著的诊断价值,95% CI为0.651~0.827,AUC值为0.739,大于0.7说明本研究Logistic回归预测模型可行性较好。选取最大正确诊断指数(Youden Index)时的最佳界值为0.325,此时敏感度为66.0%,特异度为73.1%。
The logistic regression model was used to statistically analyze the data, calculate the probability value (Pre) of fatigue prediction, and use the Pre value to analyze the ROC curve. The ROC curve of predictive fatigue was plotted according to the diagnostic probability logit(P). The results showed that the P value was 0.000, less than 0.01, indicating that Logit(P) had significant diagnostic value for fatigue, and the 95%CI was 0.651~0.827, and the AUC value was 0.739, which was greater than 0.7, indicating logistic regression in this study The feasibility of the prediction model is good. The optimal cut-off value of Youden Index was 0.325, the sensitivity was 66.0%, and the specificity was 73.1%.
图2 Logistic回归模型Pre预测疲劳的ROC曲线
Fig.2 ROC curves of Logistic regression model Pre for fatigue prediction
表3 ROC分析结果AUC对比 | ||||
预测变量 | AUC | 标准误 | p | 95% CI |
回归模型Pre | 0.739 | 0.045 | 0.000** | 0.651 ~ 0.827 |
夜班占工作总时长占比 | 0.605 | 0.054 | 0.042* | 0.499 ~ 0.712 |
* p<0.05 ** p<0.01 |
3讨论
3 Discussion
陆凌玲等研究表明疲劳与个人因素和工作环境有关[13],这与本研究结果相似,通过二元Logistic多因素回归分析对影响集中医学观察点医护人员疲劳的因素进行研究发现,夜班占工作总时长占比、是否参加体育锻炼、性别、为影响疲劳的独立影响因素(P<0.05)。其中,夜班占工作总时长影响程度最大(OR=18.614),占比每增加一个百分点时,疲劳的增加幅度为18.614倍。通过ROC曲线分析,发现夜班占工作总时长能够诊断疲劳,且诊断阈值为40.83%。说明,当夜班占工作总时长超过40.83%时,即可诊断为疲劳状态。因此集中医学观察点夜班工作时长设置时,应合理调配每个观察点医护人员数量分配,保障夜班占工作总时长低于40.83%。
Lu Lingling et al. showed that fatigue is related to personal factors and work environment [13], which is similar to the results of this study, and the impact was analyzed by binary logistic multivariate regression analysis The factors of fatigue among medical staff in centralized medical observation points showed that the proportion of night shift in the total working hours, whether they participated in physical exercise, and gender were independent influencing factors for fatigue (P<0.05). Among them, night shift accounted for the largest proportion of total working hours (OR=18.614), and when the proportion increased by one percentage point, The increase in fatigue was 18.614 times. Through the analysis of ROC curve, it was found that night shift accounted for the total working time to diagnose fatigue, and the diagnostic threshold was 40.83%. It shows that when the night shift accounts for more than 40.83% of the total working hours, fatigue can be diagnosed. Therefore, when setting the working hours of night shifts in centralized medical observation points, the number of medical staff in each observation point should be reasonably allocated to ensure that the night shift accounts for less than 40.83% of the total working hours.
对是否参加体育锻炼回归结果表明,相对于不参加体育锻炼的人员,参加体育锻炼疲劳的概率是不参加体育锻炼疲劳的2.699倍数。表明医护人员处于高强度工作状态,易产生疲劳、倦怠甚至心理焦虑状态[14] ,需要适当的休息来缓解紧张的工作。此时参加体育锻炼可能会挤压休息时间,增加身体负荷,产生运动性疲劳[15]而起到相反[16]的作用。
The regression results of whether or not to participate in physical exercise showed that the probability of fatigue from participating in physical exercise was 2.699 times that of those who did not participate in physical exercise. It indicates that medical staff are in a high-intensity working state, prone to fatigue, burnout and even psychological anxiety [14], and need appropriate rest to relieve stressful work. Participation in physical activity at this time may squeeze rest time, increase physical load, and produce exercise-induced fatigue [15] and vice versa [16].role.
性别因素研究结果发现相较于男性,女性的疲劳概率是男性的0.193倍。说明相较于女性,男性更易于疲劳。隔离点医护人员占比88.5%,为主要工作人群。其中其中男医生19人,男护士2人,女医生17人,女护士72人,可以看出男性主要在医生岗位,而女性主要在护士岗位。集中观察点护士工作繁忙[17],主要为流程性和技能性工作较多,而隔离点医生则承担隔离留观病区医学观察和病情控制主要职责,压力[18]更大,所以更易疲劳。
The results of the gender factor study found that women were 0.193 times more likely to be fatigued than men. This indicates that men are more likely to get tired than women. Medical staff in isolation points accounted for 88.5% of the total, which was the main working group. Among them, there are 19 male doctors, 2 male nurses, 17 female doctors and 72 female nurses. Nurses in the centralized observation site are busy [17], mainly for procedural and technical work, while doctors in the isolation site are mainly responsible for medical observation and disease control in the isolation observation area, which is stressful [18]. ] is bigger, so it's easier to get tired.
通过ROC验证结果对比(见表3)可以发现,单独应用夜班占比预测疲劳,正确诊断的敏感性有限,应用Logistic回归模型,联合有意义的变量,建立统计学诊断模型预测值(Pre),其P值、AUC、敏感度均优于单独应用夜班占比的ROC分析,意味着联合诊断价值比较高,与柴婕[19-21]等研究结果相似,表明应用该方法有助于提高诊断准确度。
Through the comparison of ROC verification results (see Table 3), it can be found that the sensitivity of correct diagnosis is limited when the proportion of night shift is used alone to predict fatigue. The predictive value (Pre) of the statistical diagnostic model was established, and its P value, AUC, and sensitivity were better than those of the ROC analysis of the proportion of night shifts alone, which means that the joint diagnostic value was relatively high, which was comparable to that of Chai Jie [19]. -21] and other studies, suggesting that the application of this method can help improve diagnostic accuracy.
综上所述, 本研究针对集中医学观察点医护人员疲劳logistic回归模型有效且能提高疲劳诊断效能,具有一定的预测价值。基于预测模型,医院可进行针对性预防措施。具体而言,集中医学观察点工作安排要综合考虑夜班占工作总时长占比、是否参加体育锻炼、性别因素。医院管理部门要合理安排人员调度,增加医生占比,控制夜班时长小于总工作时长的40.83%,安排充分休息时间等措施,从机制上未雨绸缪避免疲劳的发生,帮助隔离点医护人员在紧张的工作中保持充沛的精力,以高效响应基层集中医学观察的使命和职责。
In conclusion, the logistic regression model of fatigue in this study is effective and can improve the diagnostic efficiency of fatigue in centralized medical observation points, and has certain predictive value. Based on predictive models, hospitals can take targeted preventive measures. Specifically, the work arrangement of centralized medical observation sites should comprehensively consider the proportion of night shifts in the total working hours, whether they participate in physical exercise, and gender factors. The hospital management department should reasonably arrange personnel scheduling, increase the proportion of doctors, control the night shift duration to be less than 40.83% of the total working hours, arrange sufficient rest time and other measures, and take precautions to avoid fatigue from the mechanism and help medical staff in isolation pointsMaintain abundant energy in the intense work to efficiently respond to the mission and responsibilities of centralized medical observation at the grassroots level.
参考文献
References