Establishing A Predictive Model Of Nonalcoholic Fatty Liver In Nurses Based On Occupational Nurse Health Cohort Study

Objective:Describe the prevalence of non-alcoholic fatty liver disease(NAFLD)in nurses based on occupational nurse health cohort studies,analyze the independent risk factors of nurses'NALFD and establish and evaluate nurses'NAFLD risk prediction model.It provides a simple,fast and effective method for early identification of nurses'NAFLD risk factors,and provides a reference for formulating a comprehensive intervention mode to reduce nurses'NAFLD.Methods:Based on the survey of NAFLD prevalence among nurses in the occupational nurse health cohort established in this research center in 2013,a total of 2201 nurses who did not have non-alcoholic fatty liver disease in 2017 were divided into a modeling cohort and a validation cohort according to a 7:3 ratio.A prospective cohort study was conducted to investigate the independent risk factors of nonalcoholic fatty liver in nurses and to establish a risk prediction model.Statistical analysis was performed using SPSS24.0,and multivariate analysis was performed by binary logistic regression.The area under the ROC curve(AUC)and the Hosmer-Lemeshow goodness-of-fit test were used to evaluate the prediction effect of the model.Results:1.The prevalence of NAFLD among nurses increased from 9.245%to 16.781%from 2013 to 2019.2.TG,ALT,BMI?24,night shift frequency>4 times/month,poor sleep quality,department(surgery,obstetrics and gynecology,pediatrics,emergency department,ICU,operating room)were risk factors for nurses with NAFLD,ball activity Time>1h/week,sweets food frequency<1 d/week,department(outpatient)as a protective factor.3.The predictive model is Logit P=4.868+0.989×TG+0.033×ALT+2.138×BMI?24+0.531×Night shift frequency>4 times/month+1.107×Poor sleep quality-2.396×Ball activity Time>1h/week×-1.130×Sweet food frequency<1-3 times/week+(0.258×surgery+0.085×Obstetrics and Gynecology+0.047×Pediatrics-0.832×Outpatient+0.634×Emergency Department+1.109×ICU+0.437×Operating room).4.The area under the ROC curve of the nurse NAFLD prediction model is 0.863(95%CI:0.830-0.896,P<0.001).When P=0.126,the sensitivity of the prediction model is72.20%,the specificity is 85.50%,and the Hosmer-Lemeshow goodness of fit test(?~2=11.412,P=0.179),the accuracy of the model's internal prediction was 84.52%.The area under the model's external verification ROC curve was 0.858(95%CI:0.809-0.906,P<0.001),the sensitivity was 86.0%,the specificity was 70.4%,and the model prediction accuracy was 85.60%.The model prediction ability was good.Conclusion:1.The prevalence of NAFLD in nurses has continued to rise.2.The risk factors that affect the occurrence of NAFLD of nurses are TG,ALT,BMI?24,night shift frequency>4 times/month,poor sleep quality,department(surgery,obstetrics and gynecology,pediatrics,emergency department,ICU,operating room);the protective factors include ball activities>1h per week,sweets consumption<1d per week,and department(outpatient).3.The nurse NAFLD early risk prediction model established in this study has high sensitivity and specificity,good fitting,and reliable prediction effect.4.It is suggested that the comprehensive NAFLD intervention model for nurses should be formulated from aspects of diet,activity and sleep,so as to reduce the incidence of NAFLD for nurses.