| Chronic cough is defined as a disease that cough persists more than 8 weeks,with no evidence of lung disease in the chest X-ray.According to epidemiological studies in China,chronic cough is the most common reason for medical treatment in respiratory specialty and community clinics,accounting for more than one third of the daily outpatient service.Long-lasting cough is a big problem that disturbing patients’ daily life,work and study.Chronic cough is a clinical symptom with complex etiology.Common causes include:cough variant asthma(CVA),eosinophilic bronchitis(EB),upper airway cough syndrome(UACS),gastroesophageal reflux cough(GERC)and Allergic Cough(AC).In order to clarify the cause of the disease,a series of examinations such as pulmonary functional test,induced sputum,and 24-hours esophageal pH monitoring are required,which takes long time and high costs,let alone some of the examinations are invasive.Due to technical conditions and patient cooperation,chronic cough has a high rate of misdiagnosis and mistreatment,which brings heavy burden to social medical care.Clinically,chronic cough is often divided into corticosteroids responsive cough(CRC)and non corticosteroids responsive cough(NCRC),depending on the patient’s responsiveness to hormonal therapy,thereby enabling targeted examination and treatment.The China’s Guideline for Diagnosis and Treatment of Cough(2015 Edition)also emphasizes that the diagnosis of chronic cough should focus on the medical history taking,and for this purpose,it has introduced empirical treatment as an early therapy for chronic cough.However,conventional empiric therapy has a large degree of blindness,is not well-targeted,and there is also a risk of excessive medication.In particular,when oral corticosteroid treatment last for 1 week,there may be risks such as stress ulcers,increased infection rate,proliferation of tuberculosis and Osteoporosis.Therefore,how to quickly and accurately identify the causes and hormone sensitivity of chronic cough is particularly important.Discriminant analysis,a statistical analysis method,that discriminates the type of a subject based on various characteristic values of a subject under the condition of classification and determination.In the medical field,discriminant analysis is widely used in the diagnosis and classification of diseases such as cancer,cardiovascular disease,and traditional Chinese medicine.The well-known domestic scholar Professor Pixin Ran used the discriminant analysis method for the early diagnosis of COPD,and developed a discriminative software based on this,and received good results.Studies at home and abroad show that corticosteroids responsive cough and non corticosteroids responsive cough have significant differences in epidemiology,medical history,clinical manifestations,and treatment.Therefore,we use discriminant analysis methods to analyse and compare characteristics,extract relevant factors that affect hormone sensitivity in cough patients,meanwhile establish a corticosteroids responsive cough discriminant model.Then we test the reliability and accuracy of the model,in order to clarify the inherent law of hormone sensitivity recognition in patients with chronic cough.This study may lead to a simple,economical and feasible method of diagnosis of chronic cough.SectionⅠ Analysis of clinical characteristics of chronic cough with different causes.Objective: We observe the clinical characteristics of the common causes of chronic coughing,then extract the related factors affecting the sensitivity of chronic cough.Methods: Patients who were in accordance with the Guideline for Diagnosis and Treatment of Cough(2015 Edition)were enrolled from outpatient clinics of the Second Affiliated Hospital of Guangzhou Medical University and in the West District Outpatient clinic during March 1,2015 to January 1,2018.Medical history,demographic data,cough-related exposure history were collected.And pulmonary function test,induced sputum,blood routine,serum total Ig E,exhaled nitric oxide,were required to determine the cause of the patient.Patients divided into corticosteroids responsive cough group(CVA+AC+EB)and non corticosteroids responsive cough(UACS+GERC+ others).The differences in clinical characteristics between the two groups were compared.Results:1.Etiology: A total of 458 chronic cough patients were enrolled,including 303 patients with corticosteroids responsive cough(208 of them were CVA patients(68.6%),58 were EB patients(19.1%),and 37 were AC patients(12.3%)).There were 155 patients with non corticosteroids responsive cough(including 106 patients with UACS(68.3%),14 patients with GERC(9%),and 35 patients with post-infection(22.5%).2.Medical history: Compared with the NCRC group,the proportion of patients with a history of allergy in the CRC group was significantly higher(29.4% and 9.0%,P=0.000),the number of patients taking regular physical exercise was significantly increased(30.4%and 37.4% P=0.005),cough-related exposure proportion was significantly higher in the NCRC group(46.5% and 33.5%,P=0.000);The smoking rate was only statistically different between the two groups(37.4% and 30.4% P=0.745).In cough and sputum,the mean cough duration was 282.8±13.2 weeks in the CRC group and 194.4±12.5 weeks in the NCRC group.There was no significant difference between the two groups(P=0.372).The number of patients in the CRC was 140(46.2%),and number of the NCRC group was42(27.1%).There was no significant difference between the two groups(P=0.349),but the proportion of cough and sputum was higher in the NCRC group than in the CRC group(27.1% vs.21.1%,P=0.006).3.Cough symptom assessment scale: The VAS score in the CRC group was5.59±1.33 units,and the VAS score in the NCRC group was 5.82±1.22 units.There was no significant difference in the VAS score between the two groups(P=0.745).4.Blood biochemical examination: Compared with the NCRC group,the eosinophil count in the blood of the CRC group was significantly increased(2.97±4.262 and1.84±0.632,P=0.03),and there was no statistical difference among the white blood cell count,red blood cell count,and neutrophil ratio were similar in the two groups(P >0.05).5.Baseline pulmonary function: Compared with the NCRC group,the FEV1/FVC in the CRC group was lower than that in the NCRC group(36.7±42.0 and 40.6±43.6,P=0.021),while no statistically significant among the FVC,FEV1,PEF,FEF25,FEF50,and FEF75(P >0.05).The extraction and comparison of discriminating factors: Considering that the statistics of exposure history is too general and the practical operation is of little significance,the exposure history is refined and more detailed discriminating factors are extracted.According to the chronic inflammatory airway management table,the exposure history is divided into work or living environment,the history of mold contact(whether the family has the habit of causing mold),the history of mites contact(the bed sheet,the washing or drying frequency of the pillow quilt,the frequency of the air conditioning filter),the animal protein or fur contact history.There is no cockroach,etc.In the history of work or living environment,the patients in group CRC who lived in non city center were significantly lower than group NCRC(46.5% and 33.5%,P=0.006),and the time of housing in group CRC was significantly higher than that of group NCRC(46.5% and33.5%,P=0.01).In the history of mould contact,the patients with contact history of the CRC group were significantly higher than that of the NCRC group(86(28%),29(18.7%),P=0.000,P=0.000).In the history of mites contact,the patients with the history of mite contact in group CRC were significantly higher than that in group NCRC(10(11.6%),1(3.4%)and 75(87.2%),18(62.1%)P=0.000,P=0.011);in the history of animal protein or fur contact,the patients with CRC group of animal protein or fur contact history were significantly higher than those of group NCRC(13(15.1%)and 8(27.6%),P=0.000),P=0.031).Conclusion: There are some differences in part of clinical features of corticosteroids responsive cough and non corticosteroids responsive cough,but the use of either clinical feature alone can not effectively identify hormone sensitivity in patients with chronic cough,in order to more accurately identify hormone sensitivity in patients with chronic cough,a comprehensive analysis of all clinical features is needed.SectionⅡ Establishing a Primary Discriminant Model of Corticosteroids Responsive Cough Using Fisher’s discriminant analysisObjective: Establish a preliminary discriminant model of hormone-sensitive cough by using Fisher criterion.Methods: Patients who met the diagnosis criteria for the Guideline for Diagnosis and Treatment of Cough(2015 Edition)were enrolled from the Second Affiliated Hospital of Guangzhou Medical University and the West District Outpatient clinic during November,1,2015 to January 1,2018.Collecting files including: 1.general medical history data;2.clinical signs;3.blood routine test and other related indicators.Establish the model according to the discriminatory factors determined by the clinical features of two groups,then classify the discriminative factors establish the discriminant function formula.Results: Using SPSS20.0 software to establish discriminant function formula according to the selected discriminant factor.1.(1)model 1 composed of 5 factors:Canonical canonical discriminant function: Function=-0.612XA+0.7XB+1.94XC+0.881XD-0.520XE-0.589 Fischer linear classification function:Category1 = 0.473XA+1.850XB+1.850XC+2.244XD+0.652XE-5.788Category2 = 0.891XA+1.371XB+0.48XC+1.643XD+1.007XE-5.46(2),model 2 composed of 9 factors:Canonical canonical discriminant function: Function2=1.186XA+0.318XB+0.716XC+0.572XD+0.511XE+0.865XF+0.642XG+0.11XH+0.047EI-1.917 Fischer linear classification function:Category1=3.915XA+4.918XB+10.843XC+19.64XD+4.197XE+7.496XF+6.566 X G+0.123XH+0.762XI-10.329Category2=3.541XA+7.825XB+8.746XC+17.658XD+2.915XE+5.299XF+4.955 X G-0.153XH+0.643XI-7.4072.The model evaluation: model 1 surface misjudgment rate APER = 33%,cross-validation misjudgment rate CVER = 36%;Model 2 surface misjudgment rate APER = 19%,cross-validation misjudgment rate CVER = 25.5%.Model 2 was more predictive to hormone sensitivity in patients with cough whose predicted sensitivity was 90.5% and specificity was 82%.Conclusion: Fisher’s discriminant analysis was used to screen out discriminating factors helping estimating hormone sensitivity of patients with cough,so as to establish a mathematical model for discriminating hormone sensitivity.It turns out to be hopeful to originally estimate the hormone sensitivity of patients with cough in the primary hospital depending on the established model. |
PDF Full Text Request