Construction Of Prediction Models For Medication Adherence In Coronary Heart Disease Patients

OBJECTIVE:The aim of this study was to explore the relationship between coronary heart disease(CHD)patients’s medication adherence and its influencing factors,then constructing the prediction model for CHD patients’ medication adherence.METHODS:This study included two stages.The first stage:The medication adherence questionnaire item pool was built based on the 8-item Morisky medication adherence questionnaire and a lot of references;The CHD patients’ medication adherence influencing factors questionnaire was built based on World Health Organization(WHO)adherence influence factors framework and the influence factors examined by exploring CHD patients’ diseased experience,and reviewing related literatures.And the two final questionnaires were formed after two rounds of experts’ consultation.161 CHD patients were recruited by convenient sampling from three general hospitals in Baoding during 1 month period from July to August in 2017.The reliability and validity was conducted with SPSS 20.0.The statistics included:the two group t-test for independent sample,Pearson correlation,factor analysis,Cronbanch’ alpha coefficient.The second stage:This study was conducted among four general hospitals in Baoding from September to December in 2017.The CHD patients who met the inclusion and exclusion criteriawere recruited in this study.A total of 370 questionnaires were collected,and 356 questionnaires were effective.The content of investigation included:the general social-demographic characteristics,disease related characteristics,medication adherence,CHD patients medication adherence influence factors,anxiety,depression,self-efficacy for appropriate medication use,social support.The analysis was conducted with SPSS20.0 and R software.The statistics included:median,quarile range,percentage,rank sum test,Spearman correlation,multiple linear regression,regression tree and so on.RESULTS:The first stage:1.The final medication adherence included one dimension and six items.All of the correlation coefficient between every item and the total scores were more than 0.3,ranging from 0.684 to 0.833;the CR values for all items were statistically significant.The cronbanch’a was 0.820;the retest reliability is 0.866 two weeks later.The factor analysis results revealed the KMO value was 0.833,Bartlett spherical test was 490.455(P<0.001),and the cumulative variance contribution was 59.156%.2.CHD patients’ medication adherence influence factors questionnaire included 15 dimension 36 items.Except for item 2,item 10 and item 36,all of the correlation coefficient between every item and the total scores were more than 0.3,ranging from 0.478 to 0.804;Except for item 2,item 10 and item 36,the CR values for all items were statistically significant.The cronbanch’a was 0.767;the retest reliability was 0.724 two weeks later.The factor analysis results revealed the KMO value was 0.678,Bartlett spherical test was 1704.689(P<0.001),and the cumulative variance contribution was 80.750%.The second stage:1.Medication adherence:the mean of medicationadherence was 14.51.Among the all item,the score of item 4 and item 6 was lower.And 78 CHD patients’ medication adherence were good,224 were moderate and 56 were poor.2.The results revealed that the predictive performance of multiple linear regression prediction model was better than regression tree prediction model.The multiple linear regression prediction model was selected eventually.The formula for prediction model of medication adherence:Y =-1.004+1.261 ×(the place of residence)+0.970 X(work status)+0.774 X(PCI)+1.363 X(CABG)+0.298 X(self-efficacy for appropriate medication use)+0.257 X(the cognition on CHD risk factors)+0.754 X(the degree of attention on CHD)+0.527 X(medication management)-0.903 X(worry about the side effect on medication)3.Self-efficacy for appropriate medication use,the degree of attention on CHD,worry about the side effect on medication,medication management,the place of residence,the cognition on CHD risk factors,work status,PCI,CABG were predictive factors for CHD patients’medication adherence.CONCLUSIONS:1.The total level of CHD patients’ medication adherence was moderate,and only 20%patients were good.2.The prediction model:Y =-1.004+1.261 X(the place of residence)+0.970 ×(work status)+0.774 ×(PCI)+1.363 X(CABG)+0.298 X(self-efficacy for appropriate medication use)+0.257 ×(the cognition on CHD risk factors)+0.754 X(the degree of attention on CHD)+0.527 ×(medication management)-0.903 X(worry about the side effect on medication)3.The medication adherence on CHD patients who lived in rural,without job or in job,and without the PCI or CABG treatment,were poor.The higher self-efficacy for appropriate medication use,the higher cognition on CHD risk factors,the higher degree of attention on CHD,the higher medication management,the medication adherence on CHD patients were better.The more patients worried about the side effect on medication,the worse medication adherence.