Exploring The Dynamic Relationship Between Glucose And Uric Acid In Panel Data Model

ObjectivesSince the relationship between diabetes and hyperuricemia was still controversial,the panel data model was utilized to explore how the interaction between fasting blood glucose(GLU)and blood uric acid(UA)would alter over time,determine the meaningful threshold,and clarify the interaction between diabetes and hyperuricemia.In addition,the statistical method of panel data model is still blank in the field of chronic diseases.This study took the lead in introducing the panel data model to the study of the relationship between diabetes and hyperuricemia,and provide methodology reference for the research of other similar problems.Methods1.Building historical cohort of GLU and UA according the people’s physical examination data from health examination center in 2010-2013.2.To study the dynamic changes of GLU and UA in different characteristics of population.3.Using the GEE model to preliminary study the dynamic changes of the interaction between GLU and UA at different levels.4.Using the panel data model to study the dynamic changes of the interaction between GLU and UA at different levels,and determine the meaningful point.And to explore the relationship between GLU and UA in different gender,age,BMI,blood pressure,blood lipids.Results1.A total of 18275 people were collected in the four-year continuous health examination data,including 11931 males(65.3%)and 6344 females(34.7%).The average age of the surveyed population was 42.44?12.45,the women was42.09?12.04,the men was 42.63?12.66.The level of GLU within a small range and the level of UA was steady after smaller decline in four years.2.The panel data model behaved better in fitting the effect of GLU on UA,and parameter values was much better,including,R~2=0.81,F=12.83,P<0.001.The effect of GLU on UA was“bell-shaped”with threshold of 5.07 mmol/L,in other words,the GLU would promote the UA increased when it less than 5.07 mmol/L,and the effect is opposite when the GLU is lower than this value.The panel data model also behaved better in fitting the effect of UA on GLU,which parameter values include R~2=0.79,F=11.49,P<0.001.The effect of UA on GLU was“U-shaped”with threshold of 632.04?mol/L,which the UA would reduce the GLU when it less than632.04?mol/L,and the effect is opposite when the UA is lower than this value.3.The curve about the effect of UA on GLU were same in male populations of different age groups,showing a"bell-shaped".However,the threshold gradually increases with age in the male population,and the duration of the promoting effect about GLU on UA was increase.For example,GLU would promote elevated UA at GLU less than 3.94 mmol/L in the 30-40 age group,however,in the age group over70 years old,this threshold was 6.97 mmol/L.In the female population,the model parameters about the effect of UA on GLU were not statistically significant before the age of 50 years old,and the trend of change was the same as that of males after the age of 50.The curve about the effect of GLU on UA were all same in male populations of different age groups,showing a"U-shaped",and the threshold first increases and then decreased with age.In the female population,the model parameters about the effect of GLU on UA were not statistically significant before the age of 60years old,and the trend of change was the same as that of males after the age of 60.4.The curve about the effect of UA on GLU were all first rises and then falls in the people of different BMI,blood pressure and blood lipids,but the threshold in obesity group were higher than that in normal group,hypertension group was larger than normal group,and dyslipidemia group was larger than that in normal group.The curve about the effect of GLU on UA were all first falls and then rises in the people of different BMI,blood pressure and blood lipids,but the threshold in obesity group were smaller than that in normal group,hypertension group was smaller than normal group,and dyslipidemia group was smaller than that in normal group.Conclusion1.The panel data model could control individual differences and improve the accuracy of the estimation results.In this study,the result of the model fitting was well,and the panel data model is applicable to the repeated measurement of chronic disease-related indicators.2.The effect of GLU on UA in the people showed a promoting effect when the GLU was less than 5.07mmol/L,and the latter shows an inhibitory effect when the GLU was higher than this value.The effect of UA on GLU was firstly expressed as inhibitory effect and then as a promoting effect,the threshold was 632.04?mol/L.3.Age and gender are effect modifiers.With the increase of age,the promoting effect of GLU on UA increased gradually,indicating that the risk of hyperglycemia caused by GLU increased gradually.With the increase of age,the promoting effect strength of UA on GLU first strengthen and then reduction,and the minimum intensity at 60 years old,indicating that the risk of diabetes caused by UA first reduce and then increase.The relationships between GLU and UA was different in premenopausal and postmenopausal in the female people.There is no correlation between GLU and UA in premenopausal people,and the regular was the same as male after menopause.4.BMI,blood pressure,and blood lipids all enhance the effect of GLU on UA.Therefore,when obesity impairs glucose tolerance in obese,high blood pressure,and hyperlipidemia patients,UA should be monitored in time to prevent hyperuricemia.BMI,blood pressure,and blood lipids also increased the effect of blood uric acid on blood glucose.When the uric acid values reached 557.28?mol/L,560.41?mol/L,and595.53?mol/L in the people of Obesity,hypertension,and hyperlipidemia respectively,the GLU should be monitored and controlled in time.