| Objective:This study is aimed to investigate the rate of glycated hemoglobin improvement in patients with type 2 diabetes mellitus in Tianjin.Using statistical models to further explore the factors affecting glycated hemoglobin improvement,and providing a basic theory for the control of disease prevention and a comprehensive reasonable advice for health education.Comparing multivariate adaptive regression splines model(MARS)and logistic regression in the study of affecting glycated hemoglobin improvement,and evaluating whether the performance of the MARS model is better.Methods:From January 2015 to August 2017,1632 outpatients with T2DM in 19hospitals in Tianjin were recruited,the patient’s basic demographic characteristics,diabetes prevalence,and treatment variables were statistically described.The Hb A1c improvement in each group are counted,and the improvement in multi-class variables are described using a line chart to show their changing trends.The total population included in the study was divided into two groups according to Hb A1c<7%as improvement group and Hb A1c≥7%as the non-improvement group.The different number of patients in each group and their proportion were counted and tested.Fit the multivariate adaptive splines regression model to explore the complex relationship between the Hb A1c improvement and related influencing factors.At the same time,to compare the performance of the two models,logistic regression model was used to model the set,and related evaluation indicators were used to judge whether the MARS model applied to the analysis of glycated hemoglobin improvement was better.Results:Among 1,632 patients with type 2 diabetes mellitus,there were 865 male patients,accounting for 53%,and the age range was 18 to 89 years old.Among them,patients aged 60 years and over accounted for 39.09%.A total of 711 patients improved Hb A1c,with Hb A1c improvement rate of 43.57%.Among them,there were 397 patients in the male,and the rate of Hb A1c improvement was 45.90%,which was higher than that of female patients(40.94%).The rate of Hb A1c improvement in all age groups increased first and then decreased.The rate of Hb A1c improvement in the 40-50 age group was the highest,and the rate of Hb A1c improvement in the more than 60 years old group was the lowest.Patients who are younger,lower BMI,family history,shorter duration of T2DM,no related complications,insulin treatment as soon as possible,more exercise per week,no hypertension,lower fasting blood glucose,and lower 2h blood glucose after meals,their rate of Hb A1c improvement is higher(P<0.0001).The results of MARS model show that BMI lower than 22.43 kg/m~2,fasting blood glucose lower than 8.8 mmol/L,and 2h PBG lower than 10.6 mmol/L is advantageous to the Glycated hemoglobin meets standards.The interaction between age>50 years and fasting blood glucose<8.8 mmol/L is not conducive to the glycated hemoglobin improvement in patients.When the BMI is lower than 22.43 kg/m~2,the insulin doses are different(<16U,16-18U and>18U)and the effect on the glycated hemoglobin target is different,and it is not conducive to the glycated hemoglobin improvement at low doses.The performance evaluation results of the MARS model and the logistic regression model on the training set and the test set respectively show that,in the training set,the accuracy,F1 score,G-mean,and AUC of the MARS model are greater than those of the logistic regression model.The logistic regression model is slightly smaller(0.5486 vs 0.5889).In the test set,the accuracy of the MARS model,Youden index,F1 score,G-mean,and AUC were all larger than those of the logistic regression model.Conclusion:The rate of Hb A1c improvement in patients with type 2 diabetes mellitus in Tianjin is relatively low,which needs to be further improved.The results of the MARS model constructed in this study show that there is an interaction between the factors that affect the glycated hemoglobin improvement.Therefore,medical staff should take comprehensive intervention measures for patients to further control the level of glycated hemoglobin.Hierarchical management is used to control the blood glucose levels of different populations.For young,early-diagnosed patients without complications,it is recommended to receive intensive treatment as soon as possible to delay and reduce the risk of complications and improve patients’life treatment.Patients who are older,have a longer course of disease and have experienced complications,in addition to receiving conventional hypoglycemic treatment,also need to receive relevant specialist treatments to delay the progression of complications,reduce the mortality of patients,and improve the quality of life of patients.At the same time,the control of people with BMI within the normal range cannot be ignored,and the treatment must be reasonable and adequate.In addition,this article innovatively applied the multivariate adaptive regression splines(MARS)model to the analysis of influencing factors of Hb A1c improvement.Not only did it analyze the influencing factors of glycated hemoglobin improvement,but also found their interaction.According to the performance of the model,multiple indexes were used to evaluate it,and the comprehensive performance of the MARS model was better than logistic regression. |
PDF Full Text Request