A Study On Risk Assesment Model Of Zhejiang Residents’ Hearing Losses

ObjectTo explore the status and influencing factors of high-frequency hearing loss（HL）in Zhejiang residents,establish a high-frequency HL risk assessment model for residents in Zhejiang Province,and provide screening and targeted health interventions for high-risk HL population in Zhejiang Province in accordance with.MethodFrom October 2016 to January 2018,residents aged 25 years or older undergo regular physical examination in communities and hospitals in Jiangshan City,Tonglu County,Anji County,Jinyun County,Jiaxing City,and Hangzhou Municipality of Zhejiang Province.The questionnaire was designed to conduct a questionnaire survey of residents in Zhejiang Province,and at the same time,a hearing test was conducted on the residents using a professional pure-tone hearing test method.The collected data was set up using Epidata3.1.The statistical analysis method used SPSS20.0,P<0.05means the difference was statistically significant.Logistic regression was used to analyze the influencing factors of high-frequency HL and a Logistic regression model was established.The Homer-Lemeshow goodness-of-fit test was used to judge the fit of the model.The ROC curve was used to evaluate the model.Combining the results of sensitivity and specificity of each cut point,the point closest to the upper left side of the curve was used as the best critical point of the model;the model was tested using 10%of the validation samples.Results1.Basic situation4745 pieces of required information were collected through questionnaires and pure tone audiometry.After the data were seriously omitted and the information was suspiciously investigated,4385 valid questionnaires were obtained,with an effective rate of 92.4%.In the general database,high-frequency HL and normal hearing persons（normally any frequency）are combined into a new database,totaling 4037persons.90%of the samples are used to build models and 10%of samples are used to verify the effectiveness of the model.2.Influencing factors of high frequency HLThe univariate Logistic regression analysis found general demographic characteristics:age,gender,education level;family history:family genetic history;current history:tinnitus,earache,ear nausea,hypertension,hyperlipidemia,diabetes mellitus,headache,coronary heart disease,acute and chronic otitis media,chronic kidney disease;living behaviors:smoking,secondhand smoke,drinking,long sleep,earphone volume,daily consumption of a pound of fruits and vegetables,physical activity;work and living environment:work Noise,living noise,these 23 factors have statistically significant difference（P<0.05）,and may be related to high-frequency HL.Then the high-frequency single factor analysis has statistically significant factors to carry out multi-factor Logistic.Analysis to control the influence of confounding factors,and ultimately to get general demographic characteristics:age,education level;current medical history:tinnitus,ear boring,hypertension,hyperlipidemia,acute and chronic otitis media;lifestyle behavior:smoking,secondhand smoke,headset volume,physical activity;work and living environment:work noise,12factors related to the high-frequency HL.The high degree of education,physical activity,these two factors are the protective factors of high-frequency HL,the remaining 10 factors are high-risk HL risk factors.3.Establishment and verification of high frequency HL risk assessment modelLogitP=-5.028+0.095X₁-0.468X₃+0.490X₅+0.534X₇+0.710X₈+0.420X₉+1.478X₁₄+0.502X₁₇+0.201X₁₈+0.418X₂₁-0.218 X₂₃+0.271 X₂₄T=-5.028+0.095*age-0.468*literacy+0.490*tinnitus+0.534*ear stuffy+0.710*hypertension+0.420*high-lipid+1.478*acute-chronic otitis media+0.502*smoking+0.201*second hand smoke+0.418*headphone volume-0.218*physical activity+0.271*working noiseThe model obtained in the Homer-Lemeshow goodness-of-fit test was P=0.092（P>0.05,indicating good model fit）.The area of the model under the ROC curve was0.871（see Table 1）.The standard error was 0.006,P.=0.000,95%CI:（0.860-0.883）（ROC curve is shown in Figure 1）.It was found that when the sensitivity was 78.2%and the specificity was 82.5%under the ROC curve,the Youden index was 0.607 at the maximum.At this time,P=0.5662,and the cumulative total score was T=0.2664.The validation of the model effect is based on a 10%validation sample.The final positive predictive value is 80.5%,the negative predictive value is 79.3%,the compliance rate is 79.9%,and the Kappa value is 0.597.Conclusions1.Age,tinnitus,ear nausea,hypertension,hyperlipidemia,acute and chronic otitis media,smoking,secondhand smoke,earphone volume,and work noise are independent risk factors of high frequency HL.High educational level and physical activity are protective factors for high-frequency HL2.Established a high-frequency HL risk assessment model for residents in Zhejiang Province,which can be used for screening high-risk HL high-risk groups and provide evidence for targeted health management of high-risk groups.