-
Baseline characteristics of the 11, 771 participants are shown in Table 1. Details of the proportion of observations with missing information on risk factors can also be found in this table. More than 90% of the participants had complete data on BMI, HDL-cholesterol, smoking status, and creatinine. The training set and test set contributed 28, 542 and 40, 580 person-years respectively, during the follow-up time. Final events were observed in 45 participants from Huai'an city and 32 participants in Suzhou.
Table 1. Baseline Characteristics of the Participants in the Training and Test Sets
Variables Training Set (N=5, 705) Test Set (N=6, 066) Men, n (%) 2, 319 (40.65) 2, 749 (45.32) Age at diabetes diagnosis, y 55.42±9.82 55.47±9.00 Urban, n (%) 1, 596 (27.98) 3, 329 (54.88) Smoking status Current or ex-smoker, n (%) 1, 546 (27.71) 1, 775 (29.51) Nonsmoker, n (%) 4, 033 (72.29) 4, 239 (70.49) Smoking status not recorded, n (%) 126 (2.21) 52 (0.86) Clinical values BMI recorded, n (%) 5, 700 (99.01) 6, 056 (99.84) BMI (kg/m2) 25.68 (3.47) 24.69 (3.13) HDL recorded, n (%) 5, 685 (99.65) 6, 062 (99.93) HDL (mmol/L) 1.48±0.48 1.50±0.40 Creatinine (mmol/L) 73.26±27.09 72.18±25.95 Systolic blood pressure recorded, n (%) 5, 679 (99.54) 6, 017 (99.19) Systolic blood pressure, mmHg 143.79±20.45 148.42±19.58 Clinical condition Hypertension or dyslipidemia, n (%) 2, 377 (41.67) 3, 074 (50.68) Coronary heart disease, n (%) 305 (5.35) 178 (2.93) Stroke, n (%) 444 (7.78) 237 (3.91) Family history of kidney disease, n (%) 30 (0.53) 30 (0.49) Family history of T2DM, n (%) 1, 156 (20.26) 1, 699 (28.01) Family history of CVDs, n (%) 271 (4.75) 207 (3.41) Diet control or physical activity, n (%) 1, 985 (34.79) 1, 486 (24.50) Antidiabetic treatment, n (%) 4, 022 (70.50) 5, 085 (83.83) Note. Continuous data are presented as means±standard deviation; categorical data are shown as n (%). The crude incidence of DN in all patients was 9.95 cases/10, 000 person-years (95% CI 8.66-11.43) for women and 11.28 cases/10, 000 person-years (95% CI 9.77-13.03) for men. The difference in incidence between the two cohorts was not statistically significant (P=0.087).
-
Table 2 shows the variables remaining in the final risk model after forward stepwise selection. High creatinine level, hypertension, dyslipidemia, and retinopathy were all associated with increased risk of DN. The highest risk of DN occurred in men with high serum creatinine levels. Interestingly, both BMI and diet control/physical activity were significant predictors in men but not in women.
Table 2. Estimated Hazard Ratios for Risk Factors in the Final Model
Variables Hazard Ratio (95% CI) Women Men Agea 0.85 (0.76-0.92) 0.89 (0.79-0.99) BMIb n/af 0.91 (0.79-0.97) Creatininec 4.11 (2.78-6.07) 11.27 (7.67-16.56) HDL cholesterolb 0.27 (0.15-0.46) 0.28 (0.17-0.47) Locationd 0.35 (0.24-0.53) 1.78 (1.18-2.69) Hypertension or dyslipidemiae 1.52 (1.02-2.28) 1.99 (1.25-2.73) Retinopathye 5.80 (3.89-8.94) 4.00 (2.74-5.86) Diet control or physical activitye n/af 1.85 (1.25-2.73) Note. Fractional polynomial terms are defined as following: aAge/5 for both men and women; bVariables were centered in model; cln 'creatinine' for women, ln 'creatinine/100' for men. dPeople in rural areas compared with those in urban areas. eCompared with people without the condition during follow-up. fHazard ratio without statistical significance (P > 0.05). -
Using the model fashioned from the training set, the prediction in Table 3 shows the model's performance with different indicators. In men, the index values were larger. Figure 1 shows the calibration plots of our risk model. The graph shows good agreement between the observed risks and predicted risks at both 10-years and 20-years follow-up. However, a small degree of under-prediction of the estimated risk was observed at 5-years follow-up. The top quartiles of predicted risk identified 86.36% of women and 86.96% of men who developed DN over 10-years follow-up (sensitivity). The proportion of participants without DN who were not in the top quartile of the predicted risk was up to 80.41% for women and 80.67% for men (specificity).
Table 3. Model Performance with Different Indexes for Men and Women
Model R2 (95% CI) DaStatistic (95% CI) C-index (95% CI) Chi-square P Value Women Training set 75.48 (68.61-82.61) 0.82 (0.68-0.96) 0.84 (0.80-0.88) 10.45 0.11 Test set 75.00 (60.43-78.45) 0.79 (0.69-0.89) 0.79 (0.74-0.84) 11.20 0.08 Men Training set 83.10 (79.56-90.61) 0.81 (0.70-0.91) 0.80 (0.74-0.86) 14.25 0.08 Test set 86.47 (80.77-95.05) 0.89 (0.83-0.96) 0.85 (0.81-0.90) 14.56 0.07 Note. aSomers' D statistic. -
The study 'Comprehensive Research on the Prevention and Control of the Diabetes' (CRPCD) used stratified cluster sampling method in the 65 townships of three areas, in which 39, 564 T2D people had registered and received management of the 2012 National Basic Public Health Service (Changshu, Suzhou City & Huaiyin and Chuzhou, Huai'an City, Jiangsu Province, China). Townships that conducted other health care programs were excluded. Totally 29, 705 registered T2D patients in 44 townships were selected and 23, 240 individuals were recruited. Finally 20, 340 subjects undertook the questionnaire survey and physical examination (detailed consents were displayed in Table S1). All the examinations were performed within two months (from Dec, 2013 to Jan, 2014). Laboratory tests were done among 20, 053 participants respectively. The study protocol was approved by the Ethic Board of Jiangsu Provincial Center for Disease Control and Prevention (No. 2013026). All patients were well informed and signed a written consent before participating in this study.
Table Table S1. Details of Baseline Survey, Physical Examination and Laboratory Test
Baseline Survey Physical Examination Laboratory Test Demography Height HbA1c, fasting plasma glucose Lifestyle Weight Fasting lipid profile: total cholesterol, TG, HDL-C and LDL-C Tobacco and alcohol history Waist and hip circumstance Disease history Blood pressure Serum creatinine Medications and adherence Heart rate ALT and AST Family history GOT, BUN and UA Note.ALT, Alanine Transaminase; AST, Aspartate Transaminase; BUN, Blood Urea Nitrogen; GOT, Glutamic Oxaloacetic Transaminase; HbA1c, glycated hemoglobin; HDL-C, High-Density Lipoprotein Cholesterol; LDL-C, Low-Density Lipoprotein Cholesterol; TG, Triglyceride; UA, Uric Acid. Table Table S2. Definition of Diagnosis of Disease
Criteria/Variable Definition T2DM FPG≥7.0 mmol/L or previous diagnosis in medical record or having a self-reported T2DM history (ICD-9, code 250.x). Non-fatal CVDs Determined according to patients' medical records or self-report that judged by staffs. it referred to a history of coronary heart disease or stroke CHD A history of hospitalization for myocardial infarction or a surgical history of coronary balloon angioplasty, or coronary stent implantation or coronary artery bypasses (ICD-9, codes 410-414). Stroke A history of language or physical dysfunction continuing for more than 24 h and ischemic or hemorrhagic stroke diagnosed using imaging examination (ICD-9, codes 430-438). Hypertension Average blood pressure between the two measurements≥140/90 mmHg, or a previous diagnosis of hypertension. (ICD-9, codes 401-405) Dyslipidemia TG≥1.7 mmol/L or LDL-C≥3.2 mmol/L or HDL-C < 0.9 mmol/L in males, or HDL-C < 1.0 mmol/L in females Note. CHD, Coronary Heart Disease; CVD, Cardiovascular Disease; FPG, Fasting Plasma Glucose; HDL-C, High-Density Lipoprotein Cholesterol; LDL-C, Low-Density Lipoprotein Cholesterol; ICD-9, International Classification of Disease, 9th Revision; TG, Triglyceride; T2DM, Type 2 Diabetes Mellitus.
doi: 10.3967/bes2017.014
Development and Validation of a Model for Predicting Diabetic Nephropathy in Chinese People
-
Abstract:
Objective To develop a risk model for predicting later development of diabetic nephropathy (DN) in Chinese people with type 2 diabetes mellitus (T2DM) and evaluate its performance with independent validation. Methods We used data collected from the project 'Comprehensive Research on the Prevention and Control of Diabetes', which was a community-based study conducted by the Jiangsu Center for Disease Control and Prevention in 2013. A total of 11, 771 eligible participants were included in our study. The endpoint was a clear diagnosis of DN. Data was divided into two components:a training set for model development and a test set for validation. The Cox proportional hazard regression was used for survival analysis in men and women. The model's performance was evaluated by discrimination and calibration. Results The incidence (cases per 10, 000 person-years) of DN was 9.95 (95% CI; 8.66-11.43) in women and 11.28 (95% CI; 9.77-13.03) in men. Factors including diagnosis age, location, body mass index, high-density-lipoprotein cholesterol, creatinine, hypertension, dyslipidemia, retinopathy, diet control, and physical activity were significant in the final model. The model showed high discrimination and good calibration. Conclusion The risk model for predicting DN in people with T2DM can be used in clinical practice for improving the quality of risk management and intervention. -
Table 1. Baseline Characteristics of the Participants in the Training and Test Sets
Variables Training Set (N=5, 705) Test Set (N=6, 066) Men, n (%) 2, 319 (40.65) 2, 749 (45.32) Age at diabetes diagnosis, y 55.42±9.82 55.47±9.00 Urban, n (%) 1, 596 (27.98) 3, 329 (54.88) Smoking status Current or ex-smoker, n (%) 1, 546 (27.71) 1, 775 (29.51) Nonsmoker, n (%) 4, 033 (72.29) 4, 239 (70.49) Smoking status not recorded, n (%) 126 (2.21) 52 (0.86) Clinical values BMI recorded, n (%) 5, 700 (99.01) 6, 056 (99.84) BMI (kg/m2) 25.68 (3.47) 24.69 (3.13) HDL recorded, n (%) 5, 685 (99.65) 6, 062 (99.93) HDL (mmol/L) 1.48±0.48 1.50±0.40 Creatinine (mmol/L) 73.26±27.09 72.18±25.95 Systolic blood pressure recorded, n (%) 5, 679 (99.54) 6, 017 (99.19) Systolic blood pressure, mmHg 143.79±20.45 148.42±19.58 Clinical condition Hypertension or dyslipidemia, n (%) 2, 377 (41.67) 3, 074 (50.68) Coronary heart disease, n (%) 305 (5.35) 178 (2.93) Stroke, n (%) 444 (7.78) 237 (3.91) Family history of kidney disease, n (%) 30 (0.53) 30 (0.49) Family history of T2DM, n (%) 1, 156 (20.26) 1, 699 (28.01) Family history of CVDs, n (%) 271 (4.75) 207 (3.41) Diet control or physical activity, n (%) 1, 985 (34.79) 1, 486 (24.50) Antidiabetic treatment, n (%) 4, 022 (70.50) 5, 085 (83.83) Note. Continuous data are presented as means±standard deviation; categorical data are shown as n (%). Table 2. Estimated Hazard Ratios for Risk Factors in the Final Model
Variables Hazard Ratio (95% CI) Women Men Agea 0.85 (0.76-0.92) 0.89 (0.79-0.99) BMIb n/af 0.91 (0.79-0.97) Creatininec 4.11 (2.78-6.07) 11.27 (7.67-16.56) HDL cholesterolb 0.27 (0.15-0.46) 0.28 (0.17-0.47) Locationd 0.35 (0.24-0.53) 1.78 (1.18-2.69) Hypertension or dyslipidemiae 1.52 (1.02-2.28) 1.99 (1.25-2.73) Retinopathye 5.80 (3.89-8.94) 4.00 (2.74-5.86) Diet control or physical activitye n/af 1.85 (1.25-2.73) Note. Fractional polynomial terms are defined as following: aAge/5 for both men and women; bVariables were centered in model; cln 'creatinine' for women, ln 'creatinine/100' for men. dPeople in rural areas compared with those in urban areas. eCompared with people without the condition during follow-up. fHazard ratio without statistical significance (P > 0.05). Table 3. Model Performance with Different Indexes for Men and Women
Model R2 (95% CI) DaStatistic (95% CI) C-index (95% CI) Chi-square P Value Women Training set 75.48 (68.61-82.61) 0.82 (0.68-0.96) 0.84 (0.80-0.88) 10.45 0.11 Test set 75.00 (60.43-78.45) 0.79 (0.69-0.89) 0.79 (0.74-0.84) 11.20 0.08 Men Training set 83.10 (79.56-90.61) 0.81 (0.70-0.91) 0.80 (0.74-0.86) 14.25 0.08 Test set 86.47 (80.77-95.05) 0.89 (0.83-0.96) 0.85 (0.81-0.90) 14.56 0.07 Note. aSomers' D statistic. Table S1. Details of Baseline Survey, Physical Examination and Laboratory Test
Baseline Survey Physical Examination Laboratory Test Demography Height HbA1c, fasting plasma glucose Lifestyle Weight Fasting lipid profile: total cholesterol, TG, HDL-C and LDL-C Tobacco and alcohol history Waist and hip circumstance Disease history Blood pressure Serum creatinine Medications and adherence Heart rate ALT and AST Family history GOT, BUN and UA Note.ALT, Alanine Transaminase; AST, Aspartate Transaminase; BUN, Blood Urea Nitrogen; GOT, Glutamic Oxaloacetic Transaminase; HbA1c, glycated hemoglobin; HDL-C, High-Density Lipoprotein Cholesterol; LDL-C, Low-Density Lipoprotein Cholesterol; TG, Triglyceride; UA, Uric Acid. Table S2. Definition of Diagnosis of Disease
Criteria/Variable Definition T2DM FPG≥7.0 mmol/L or previous diagnosis in medical record or having a self-reported T2DM history (ICD-9, code 250.x). Non-fatal CVDs Determined according to patients' medical records or self-report that judged by staffs. it referred to a history of coronary heart disease or stroke CHD A history of hospitalization for myocardial infarction or a surgical history of coronary balloon angioplasty, or coronary stent implantation or coronary artery bypasses (ICD-9, codes 410-414). Stroke A history of language or physical dysfunction continuing for more than 24 h and ischemic or hemorrhagic stroke diagnosed using imaging examination (ICD-9, codes 430-438). Hypertension Average blood pressure between the two measurements≥140/90 mmHg, or a previous diagnosis of hypertension. (ICD-9, codes 401-405) Dyslipidemia TG≥1.7 mmol/L or LDL-C≥3.2 mmol/L or HDL-C < 0.9 mmol/L in males, or HDL-C < 1.0 mmol/L in females Note. CHD, Coronary Heart Disease; CVD, Cardiovascular Disease; FPG, Fasting Plasma Glucose; HDL-C, High-Density Lipoprotein Cholesterol; LDL-C, Low-Density Lipoprotein Cholesterol; ICD-9, International Classification of Disease, 9th Revision; TG, Triglyceride; T2DM, Type 2 Diabetes Mellitus. -
[1] Hovind P, Rossing P, Tarnow L, et al. Progression of diabetic nephropathy. Kidney Int, 2001; 59, 702-9. doi: 10.1046/j.1523-1755.2001.059002702.x [2] Rossing K, Christensen PK, Hovind P, et al. Progression of nephropathy in type 2 diabetic patients. Kidney Int, 2004; 66, 1596-605. doi: 10.1111/j.1523-1755.2004.00925.x [3] Ⅱ. Incidence and prevalence of ESRD. Am J Kidney Dis, 1999; 34, S40-50. [4] Burden AC, McNally PG, Feehally J, et al. Increased incidence of end-stage renal failure secondary to diabetes mellitus in Asian ethnic groups in the United Kingdom. Diabet Med, 1992; 9, 641-5. doi: 10.1111/dme.1992.9.issue-7 [5] Hallan SI, Dahl K, Oien CM, et al. Screening strategies for chronic kidney disease in the general population:follow-up of cross sectional health survey. BMJ, 2006; 333, 1047. doi: 10.1136/bmj.39001.657755.BE [6] Sarnak MJ, Levey AS, Schoolwerth AC, et al. Kidney disease as a risk factor for development of cardiovascular disease:a statement from the American Heart Association Councils on Kidney in Cardiovascular Disease, High Blood Pressure Research, Clinical Cardiology, and Epidemiology and Prevention. Hypertension, 2003; 42, 1050-65. doi: 10.1161/01.HYP.0000102971.85504.7c [7] Fabre J, Balant LP, Dayer PG, et al. The kidney in maturity onset diabetes mellitus:a clinical study of 510 patients. Kidney Int, 1982; 21, 730-8. doi: 10.1038/ki.1982.90 [8] Haffner SM, Lehto S, Ronnemaa T, et al. Mortality from coronary heart disease in subjects with type 2 diabetes and in nondiabetic subjects with and without prior myocardial infarction. N Engl J Med, 1998; 339, 229-34. doi: 10.1056/NEJM199807233390404 [9] Tonelli M, Muntner P, Lloyd A, et al. Risk of coronary events in people with chronic kidney disease compared with those with diabetes:a population-level cohort study. Lancet, 2012; 380, 807-14. doi: 10.1016/S0140-6736(12)60572-8 [10] Gerstein HC, Mann JF, Yi Q, et al. Albuminuria and risk of cardiovascular events, death, and heart failure in diabetic and nondiabetic individuals. JAMA, 2001; 286, 421-6. doi: 10.1001/jama.286.4.421 [11] Yokoyama H, Tomonaga O, Hirayama M, et al. Predictors of the progression of diabetic nephropathy and the beneficial effect of angiotensin-converting enzyme inhibitors in NIDDM patients. Diabetologia, 1997; 40, 405-11. doi: 10.1007/s001250050694 [12] Ueda H, Ishimura E, Shoji T, et al. Factors affecting progression of renal failure in patients with type 2 diabetes. Diabetes Care, 2003; 26, 1530-4. doi: 10.2337/diacare.26.5.1530 [13] Clarke PM, Gray AM, Briggs A, et al. A model to estimate the lifetime health outcomes of patients with type 2 diabetes:the United Kingdom Prospective Diabetes Study (UKPDS) Outcomes Model (UKPDS no. 68). Diabetologia, 2004; 47, 1747-59. doi: 10.1007/s00125-004-1527-z [14] Lu B, Song X, Dong X, et al. High prevalence of chronic kidney disease in population-based patients diagnosed with type 2 diabetes in downtown Shanghai. J Diabetes Complications, 2008; 22, 96-103. doi: 10.1016/j.jdiacomp.2007.08.001 [15] Yang XL, So WY, Kong AP, et al. End-stage renal disease risk equations for Hong Kong Chinese patients with type 2 diabetes:Hong Kong Diabetes Registry. Diabetologia, 2006; 49, 2299-308. doi: 10.1007/s00125-006-0376-3 [16] Rubin DB. Multiple Imputation for Nonresponse in Surveys, New York:Wiley. Biometrics, 1989; 45. [17] Royston P, Ambler G, Sauerbrei W. The use of fractional polynomials to model continuous risk variables in epidemiology. Int J Epidemiol, 1999; 28, 964-74. doi: 10.1093/ije/28.5.964 [18] Schemper M, Henderson R. Predictive accuracy and explained variation in Cox regression. Biometrics, 2000; 56, 249-55. doi: 10.1111/j.0006-341X.2000.00249.x [19] Heagerty PJ, Zheng Y. Survival model predictive accuracy and ROC curves. Biometrics, 2005; 61, 92-105. doi: 10.1111/biom.2005.61.issue-1 [20] Royston P. Explained variation for survival models. Stata Journal, 2006; 6, 83-96. http://artax.karlin.mff.cuni.cz/r-help/library/pec/html/R2.html [21] Royston P, Sauerbrei W. A new measure of prognostic separation in survival data. Stat Med, 2004; 23, 723-48. doi: 10.1002/(ISSN)1097-0258 [22] Statacorp LP. Stata multiple-imputation reference manual. Stata Press Publication, 2011. [23] Manttari M, Tiula E, Alikoski T, et al. Effects of hypertension and dyslipidemia on the decline in renal function. Hypertension, 1995; 26, 670-5. doi: 10.1161/01.HYP.26.4.670 [24] Thomas GN, Critchley JA, Tomlinson B, et al. Peripheral vascular disease in Type 2 diabetic Chinese patients:associations with metabolic indices, concomitant vascular disease and genetic factors. Diabet Med, 2003; 20, 988-95. doi: 10.1046/j.1464-5491.2003.01046.x [25] Keane WF, Brenner BM, de Zeeuw D, et al. The risk of developing end-stage renal disease in patients with type 2 diabetes and nephropathy:the RENAAL study. Kidney Int, 2003; 63, 1499-507. doi: 10.1046/j.1523-1755.2003.00885.x [26] Sawicki PT, Didjurgeit U, Muhlhauser I, et al. Smoking is associated with progression of diabetic nephropathy. Diabetes Care, 1994; 17, 126-31. doi: 10.2337/diacare.17.2.126 [27] Bakris GL, Williams M, Dworkin L, et al. Preserving renal function in adults with hypertension and diabetes:a consensus approach. National Kidney Foundation Hypertension and Diabetes Executive Committees Working Group. Am J Kidney Dis, 2000; 36, 646-61. doi: 10.1053/ajkd.2000.16225 [28] Schrier RW, Estacio RO, Esler A, et al. Effects of aggressive blood pressure control in normotensive type 2 diabetic patients on albuminuria, retinopathy and strokes. Kidney Int, 2002; 61, 1086-97. doi: 10.1046/j.1523-1755.2002.00213.x [29] Tight blood pressure control and risk of macrovascular and microvascular complications in type 2 diabetes:UKPDS 38. UK Prospective Diabetes Study Group. BMJ, 1998; 317, 703-13. doi: 10.1136/bmj.317.7160.703