| Background and objectives Diabetes has become a major public health problem that threatens human health and quality of life both at home and abroad. The prevalence and distribution of risk factors for cadres in the army is different form general population. However, there was lack of the basic data for the prevalence, control status and incidence of diabetes among middle aged and elderly cadres in the army. Therefore,the aim of this study is to analysis the prevalence, control status and its changing trend for the past 5 years, and also explores the related risk factors; calculate the incidence and construct diabetesrisk model.Methods The study consists of two parts. The first part is a cross-sectional study.We used the physical examination database of 2009 and 2013 (about 50, 000 participants), to analysis of the prevalence and control status of diabetes, and also explore its 5 years' change for middle aged and elderly cadres in the army. Secondly, a cohort study design based on 24200 participants was used to analysis the incidence of diabetes, and to explore the related risk factors. 4 kinds of statistical analysis model were used to contrast the diabetes incidence risk model, and prediction efficiency was comparedResults The prevalence of diabetes in the study was 20.6% and 20.5% for the two survey years respectively. For those have stayed in both two survey years, the prevalence of diabetes had increased from 21.4% in 2009 to (p<0.001), which reflected the actual increaseof diabetes. The prevalence of diabetes increased with age, and male participants had a higher prevalence than that of female. A multi-logistic regression analysis showed that independent variables including age, BMI, SBP, HDL-C, FPG,2hPG, BUN, TBIL, gender, retirement, had statistically significant effects on the prevalence of diabetes mellitus (p<0.05). The control rate was about 50% and increased gradually with the age group (ptrend<0.05).Female had a higher control rate than male participants (p<0.05).The incidence rate of diabetes of this study population was 4.27%, while male was higher than female (4.47% vs. 2.04%, p<0.05), and increased with age group(ptrend<0.001). Age, BMI,SBP, FPG, 2hPG,ALT were positively correlated with the incidence of diabetes (RR>1), while HDL-C was negatively correlated (RR<1). For the full-population cohort with a large sample size, all the four statistical methods used to build diabetes incidence risk models had good prediction efficiency, with AUC between 0.770-0.780. However, the variables included and the coefficients were different.For the high-risk cohort with relatively small sample size, the Cox model had the best prediction ability (AUC=0.753). In the remaining three models, the AUC of the traditional Logistic regression model was lowest (AUC=0.658) and the AUC of the multilevel model and the GEE model were 0.685 and 0.683 respectively.Conclusions The prevalence of diabetes among middle aged and elderly cadres in the army in the study increased has increased over the past 5 years, but was still lower than the general population. Multiple factors were associated with diabetes. The control rate of diabetes in the study was relatively better than general population. The diabetes incidence of this study population rate is 4.27%which was lower than the same age group in the general population. For the full-population cohort with a large sample size,all the four statistical methods used to build diabetes incidence risk models had good prediction efficiency, even with different predictive ability and standard errors. For the high-risk cohort with relatively small sample size, the Cox model had the best prediction ability when we have time information while traditional Logistic model had the lowest AUC. |
PDF Full Text Request