| BackgroundCharacterized by hyperglycemia, Diabetes mellitus （DM） is an endocrine and metabolic disease, caused by multiple pathogenic factors such as genetic, environmental and behavioral ones. According to statistics, more than 90% DM are type 2 diabetes mellitus （T2DM）. Impaired glucose regulation （IGR）, also called prediabetes （PDM）, is a transition period between normal glucose tolerance （NGT） and DM.The prevalence of T2DM is increasing in the world and developing into a momentous global public health problem in the 21 century for many reasons like rapid economic growth, social development, urbanization, and increase in life expectancy and changes in life style. More and more people like less activity, longtime sedentary work and high calorie diet. T2DM in China is also the case. According to the last statistic of International Diabetes Federation （IDF）, it would appear China has overtaken India and become the global centre of the diabetes epidemic with 92.4 million adult diabetics. The burden of diabetes is mainly from its complications. According to the "China Health Statistical Yearbook （1993～2004）", the annual growth rate of the direct medical cost of diabetes reached 19.90%, over the rate of growth of the GDP and the total health expenses of the same period, occupying the second position in all the chronic disease investigated. T2DM and its complications have brought grave economic burden and social pressure to China.What makes us so worried is that the rate of undiagnosed diabetes is more than 50% in the research subjects. As most T2DM has no specific symptoms in early stage, there’s a latent period of 9 to 12 years before clinical diagnosis. Many patients have diabetes complications in heart, brain, kidney, eye and other organs when final diagnosis is made, which not only seriously affect the patient’s physical and mental health as well as quality of life, but also bring heavy burden to the individuals, families and society.The prevalence of IGR is high as well, which has greatly exceeded the prevalence of diabetes. The results of several large prospective studies published from 1997 to 2006 for intervention to the individuals with high risk of diabetes showed that early detection and intervention for IGR by life style promotion and drug treatment will reduce about 50% risk to worsen into diabetes.In brief, in view of the great harm and heavy burden caused by the high prevalence of T2DM and IGR and the evidence from several large-scale randomized controlled trials, it’s necessary for us to identify diabetes and high risk individuals earlier by screening. Early detection and intervention can not only promote half of them to recover to the normal glucose but also can prevent the occurrences of diabetes and delay its complications. Therefore, it is of great importance for us start diabetes screening programs to curb the high prevalence of diabetes in China.Based on the risk factors of T2DM, we can identify suspicious patients and high risk individuals. Purposeful early detection and focused intervention for T2DM and IGR in the targeted population instead of general population is the key strategy for diabetes prevention. Studies indicate that establishing diabetes risk assessment models can make the total target population reduced greatly, which can improve the detection efficiency and save the cost of screening. Several countries such as Finland, UK, USA, Germany, Australia and India, have undertook study in the domain of T2DM risk assessment and prediction, and obtained a lot of achievements and experience. But because of racial differences, further research is needed to verify them in Chinese population. Although there has been several large investigation of diabetes prevalence in our country, but the research of community-based screening methods and strategies for T2DM and prediabetes is relatively backward. There is less comprehensive and through research on the selection of the targeted population, development of screening methods and strategies, evaluation of joint use of the methods and the influence of selection of the cut-off values on the screening effect. And there are no generally accepted diabetes risk scores or self-administered questionnaires that are simple, economic and effective for community-based screening for T2DM and IGR in China.Objectives1. To construct optimal screening models appropriate to identify asymptomatic T2DM and T2DM/IGR individuals among the northern Chinese population. On the basis of the best models, a T2DM risks self-assessment system is developed. It is scientific and effective and can be used as a simple and practical first-line tool to detect T2DM and T2DM/IGR in communities. The specific goals include:a) To describe the prevalence of T2DM and IGR and analyze their related risk factors—taking Qingdao survey in 2006 as a case study;b) To evaluate comprehensively the screening effects of every single indicator selected and their combinations for T2DM and T2DM/IGR;c) To construct a screening indicators system and optimal models for T2DM and T2DM/IGR.2. To present screening strategies for community asymptomatic T2DM and IGR in northern Chinese population and select an effective, simple and practical community screening path so as to provide reference and appropriate technology for T2DM and T2DM/IGR community diagnosis and intervention strategies.Methods1. Description of the prevalence of T2DM and IGR in Qingdao based on the survey data of 2006 Qingdao diabetes prevention program （WDF-05-108）, and analysis associated risk factors of T2DM and IGR.2. ROC analysis of 5 screening methods for T2DM and IGR:4 laboratory ones are fasting capillary blood glucose （FCBG）, fasting plasma glucose （FPG）, glucose load 2H plasma glucose （2HPG）, glycosylated hemoglobin （HbAlc） and 1 noninvasive one is the diabetes risk factors （DRS） used in 2006 Qingdao diabetes prevention program, to evaluate the effects of 5 screening methods as well as the detection values of the DRS.3. The classification and combinations and the screening indicators.a) We select the significant risk factors as the first screening indicators, such as age, body mass index （BMI）, waist circumference, blood pressure, DM family history and pulse, which are also called noninvasive screening indicators. Choosing different number of screening indicators to form various combinations with Exhaustive Method, and we get 512 combinations in total.b) We select laboratory tests as the second screening indicators, such as FCBG, FPG,2HPG, HbAlc, TG, TC, HDL, LDL, and form 52 combinations in total.c) We combined the best noninvasive model with the laboratory tests and form 20 screening models in total.4. Evaluation of the screening indicator combinations. The study uses the famous Finland diabetes risk factors （FINDRISC） as a model to calculated diabetes probability P （0<P<1） for combinations of different risk factors （screening indicators）, and according to the diagnostic gold standard, the sensitivity and specificity of every P can be computed and ROC curves constructed to evaluate the screening effects of all the multiple indicators combinations listed before.5. The optimal noninvasive screening models and the optimal combinations of laboratory indicators for T2DM and T2DM/IGR are selected according to the area under ROC （AUROC）.6. The best combined screening models are constructed based on the optimal noninvasive screening models and the optimal laboratory screening combinations for T2DM and T2DM/IGR. 7. Validation of the optimal screening models:a) The survey data of Jiaonan in 2009 is used to validate the optimal noninvasive screening models and the best combined screening models for T2DM and T2DM/IGR.b) The survey data of Jinan in 2007-2008 is used to validate the optimal noninvasive screening models and the best combined screening models for T2DM and T2DM/IGR.8. Qingdao DRS is used as a case study to analyze the influence of covariant factors on the screening model to evaluate its applicability and effects in detecting T2DM in communities.9. Comprehensive evaluation on the screening methods and strategies for asymptomatic T2DM and T2DM/IGR in communities.Results1. There is an accelerating prevalence rate of T2DM and IGR in Qingdao. The prevalence rate of T2DM was 21.8%（13.2% newly diagnosed and 8.6% known） in the subjects; 1/4 of subjects are IGR; the diabetes awareness rate was only 50%; more than 64% people did not do any physical exercise during the last 12 months. The average BMI and average waist circumstance reaches the cut-off value of overweight, and the average WHR is close to the cut-off value of central obesity both in male and female. In male, the average blood pressure including systolic and diastolic blood pressure is in the near critical value of hypertension. The mean fasting plasma glucose （FPG） is close to the critical values of IFG （6.1mmol/L） （WHO 1999 Standard） both in male and female. The mean OGTT 2 hour plasma glucose （2HPG） is 7.15 mmol/L in male while 7.43 mmol/L in female, which are close to the cut-off value of IGT （7.8 mmol/L）.2. Risk factors of T2DM and IGR in Qingdao include age increasing, hypertension, overweight or obesity, history of hypertension, high pulse rate and family history of diabetes. Regular physical activity and high degree education are protective factors.3. In five screening methods:the results of FCBG, FPG, 2HPG, HbA1c and DRS to screening asymptomatic T2DM:the screening effect of 2HPG and FPG are the best, significantly better than the other three methods. There is no significant difference between the screening result of FCBG, HbAlc and DRS, but DRS is a noninvasive screening questionnaire only including three risk factors of age, waist circumference and a family history. So the DRS is a simple, easy and effective method for Chinese population of asymptomatic diabetic screening. FCBG can be also used for screening for diabetes. HbAlC is not suitable for screening for T2DM and IGR individuals in communities at present.4. The choice of screening cut-point influence the effect of screening:the optimal screening cut-off values for T2DM, FCBG, FPG,2HPG, HbAlc are>6.5,6.7,9.4 mmol/L and 5.5% respectively; to screening for IGR, the optimal cut-points are 5.6,5.6,7.0 mmol/L and 5.21% respectively; to screening for T2DM/IGR, the optimal cut-points are>5.6,5.6,7.0mmol/L and 5.27% respectively.5. The best noninvasive risk assessment model for screening asymptomatic T2DM in communities include 10 indicators:age, gender, family history of diabetes, waist circumference, BMI, systolic blood pressure, pulse rate, history of hypertension, level of education and regular physical exercise. When the optimal cut-off value is 0.0942, the sensitivity, specificity and the ROC area under curve （AUROC） of the model are 68.8%,63.7% and 0.714 respectively.6. The best noninvasive risk assessment models for screening for T2DM/IGR in communities include 8 indicators:age, gender, family history of diabetes, waist circumference, BMI, systolic blood pressure, pulse rate and level of education. When the optimal cut-point is>0.4113, the sensitivity, specificity and AUROC are 63.6%, 67.2% and 0.711 respectively.7. The laboratory indicator 2HPG combined the best noninvasive risk assessment model has the best effect to screen asymptomatic T2DM and T2DM/IGR in communities. The sensitivity, specificity and AUROCR of the combined screening model for asymptomatic T2DM are 78.9%,92.4% and 0.91 respectively, when the optimal cut-off value is> 0.1423. To screen for T2DM/IGR, the sensitivity, specificity and ROC area under curve of the combined model are 71.4%,96.5% and 0.884 respectively, when the optimal cut-off value is> 0.599.8. Based on the Youden index, the best noninvasive screening model （1107） for T2DM has similar effect when it is validated by the survey data of Jiaonan and Jinan. The best combined screening model （C3） for T2DM has slightly lower effect when it is used in the Jiaonan while it has better effect in Jinan. Overall, the two screening optimal models for T2DM can be used in the northern area outside Qingdao to screen for asymptomatic T2DM in the communities.9. Based on the Youden index, the screening validation effect of the optimal noninvasive screening model （1822） and the best combined screening model （C12） for T2DM/IGR is slightly worse while they have a better screening effect when they are used in Jinan population. Overall, the two screening optimal models for T2DM/IGR can be used in the northern area outside Qingdao to screen for the asymptomatic T2DM/IGR in the communities.10. When screening T2DM by DRS, the detection rate of low age group outperformed higher age group, while women group outperformed men group.Conclusions and suggestions1 The noninvasive screening models based on the T2DM and IGR risk factors are simple, effective, practical first-line methods for screening asymptomatic T2DM and prediabetes in communities as well as a practical tool for diabetes risk self-assessment and health education.2 The two-step screening method combined the best noninvasive screening model with the postprandial or 75g OGTT 2-hour blood glucose is the best path applied to the Chinese northern populations for screening T2DM and IGR in communities.3 In order to improve the applicability of the screening models, when we construct and use them, we should take notice that different screening cut-off values would influence their detection effects and some covariate factors, such as age, gender would influence their sensitivity and specificity for identifying T2DM and IGR individuals. 4 In view of high prevalence of T2DM and IGR in China, we advice all the individuals to use the noninvasive screening models as a self-assessment tool to find their diabetes risk factors to raise the public and individual awareness of preventing diabetes in the communities, especially in communities with the high incidence of diabetes. If condition permits, it is necessary for community health workers perform opportunistic screening for asymptomatic T2DM and high risk individuals in the residents using stepwise strategies, i.e. to find the high risk ones by the noninvasive screening models followed by OGTT 2HPG for further detection. It’s an auxiliary means of community diagnosis leading to early identification and intervention for asymptomatic T2DM and IGR individuals.5 We advice that the government should increase investment to provide funding and technical support on the research and development of screening tools and appropriate technology for T2DM and IGR in communities. On the basis of community screening and diagnosis, large randomized clinical trials should be purposefully organized and follow-up cohort data should be collected by use of electronic health records to construct Chinese diabetes screening indicators system, surveillance mechanism and prediction models, and evaluate them to develop appropriate screening strategies in communities. It is of great significance and practical value to curb accelerating prevalence of diabetes in China and mitigate its harm and burden of disease.Innovations and limitationsThe innovations include that:1. The content is practical and creative. There is no systematic study screening indicator system and screening strategies for asymptomatic T2DM and IGR in northern China’s populations, while at present there is no well-accepted community screening tool and appropriate technology in China. Based on the analysis of diabetes prevalence in Qingdao, this study constructs a screening indicator system and evaluates all possible combinations of the indicators to select the best noninvasive screening models and validates them by the survey data from Jiaonan and Jinan. It’s the first time to carry out a comprehensive study on T2DM and T2DM/IGR screening models and operation path. On the basis of evaluation of cut-off selection and adaptability of noninvasive screening model, we try to develop suitable first-line screening tools and optimal detection strategies for diabetes in the population of North China. It is of practical value and great significance to curb the rapid and high prevalence of diabetes in China.2. The research methods are nontrivial. This study adopt a new analysis strategy, listing all the possible combinations of screening indicators through Exhaustive Method and evaluating them to select the optimal ones by the way of multivariable linear combination of ROC analysis. The method changes all possible combinations of screening indicators into incidence probability p of T2DM and T2DM/IGR, and then we can evaluate the screening effect of the combinations through computing the sensitivity, specificity and AUROC of the P values of all the subjects investigated, "descending dimension" and solving the problem of evaluating combination of multiple indicators. The new method can not only make up for the defects of the traditional analysis ones, fully mining the inherent information of the data, but also provide methodology reference for the other similar studies.3. The results are innovative. In this study, we find that noninvasive screening models are simple, effective, economic and practical first-line methods to identify asymptomatic T2DM and T2DM/IGR in communities. We can develop the noninvasive screening models into diabetes risk self-assessment software system, which will automatically calculate the risk value of diabetes when related indicators are input, and propose for further recommendations on diagnosis and health intervention. The system is few to see in China at present, which accumulates experience to the similar researches. At the same time, the best screening path which combines the noninvasive screening models with the laboratory indicator is an appropriate community technology to screen for T2DM and T2DM/IGR. It could provide evidence and guidance to formulate effective policies and strategies for diabetes prevention and control in communities. The limitations include that:1. The data for this study is from a cross-sectional survey, not a cohort research, so the screening models we developed can be only used for screening T2DM and IGR, but not for predicting future incidence of diabetes. The models can also be used as surveillance tools for diabetes risk assessment and health education and promotion. The prediction models need follow-up cohort data, but at present our country has not carried out a large-scale clinical trial and lack relevant information which is our direction of the coming research.2. The study has limitations in the selection of research subjects, which are only come from Qindao. The best noninvasive models and combined ones with laboratory indicator 2HPG for screening asymptomatic T2DM and high-risk individuals are only validated by the data from Jiaonan and Jinan, two areas in Shandong so theirs screening effectiveness requires more evaluation and improvement based on the nationwide population data. |
PDF Full Text Request