Potential medication safety risks and Lasso-logistic regression analysis in hypertension complicated with cerebral infarction elderly patients
AIM To investigate the potentially inappropriate medications(PIM)in elderly patients with hypertension combined with cerebral infarction and to develop a nomogram model based on Lasso-logistic regression of the influencing factors.METHODS A total of 105 patients with hypertension combined with cerebral infarction and multi-drug therapy discharged from January to December 2020 were included,and the PIM was evaluated by the criteria of potentially inappropriate medications for older adults in China.The PIM influencing factors were analyzed by Lasso-logistic regression,Nomogram model of factors influencing the PIM risk was developed,and the discrimination and accuracy of the model were evaluated.RESULTS The incidence of PIM in 105 patients of hypertension complicated with cerebral infarction was 85 cases(81.0％)and 184 times,involving 26 drugs.The 1st drug was clopidogrel with 48 times(26.1％).A statistically significant effect of the number of drugs(P=0.003,OR=1.412,95％CI:0.582 to 2.868)and diabetes(P＜0.001,OR=0.081,95％CI:-3.899 to-1.129)on PIM was suggested by the Lasso-logistic regression.Of the model,a good accuracy and discrimination was suggested by the area under the ROC curve(AUC),which was 85.3％(95％CI:0.772 to 0.933),and an acceptable predictive power was suggested by internal validation with Bootstrap method(B=1 000).CONCLUSION The nomogram model based on Lasso-logistic regression can better identify the risk of PIM.Elderly patients with hypertension complicated with cerebral infarction and multi-drug therapy have a high risk of PIM,and it is necessary to reduce the number of medications to reduce the risk of PIM.
NETL
NSTL
万方数据
