Age-related Associations of Hypertension and Diabetes Mellitus with Chronic Kidney Disease

Tareq M Islam; Caroline S Fox; Devin Mann; Paul Muntner


BMC Nephrology 

In This Article


NHANES 1999–2004 is a cross-sectional nationally representative survey of the non-institutionalized civilian population of the United States.[7] Briefly, this study employs a multi-stage stratified probability sample based on selection of counties, blocks, households, and persons within households. Mexican-Americans, non-Hispanic blacks, and older adults were over-sampled in order to improve the precision of estimates for these groups. NHANES 1999–2004 consisted of an in-home interview and a medical evaluation and blood sample collection in a mobile examination center. Overall, 15,332 participants completed the medical evaluation and study interview. Participants without serum creatinine measurements, estimated GFR (eGFR) < 15 ml/min/1.73 m2, and pregnant women were excluded from the analysis of stage 3 or 4 CKD, resulting in a sample size of 12,518 participants. For the analysis of albuminuria, individuals missing urinary albumin or urinary creatinine measurements and pregnant or menstruating women were excluded, resulting in a total sample size of 12,778 participants. Women and non-Hispanic blacks were more likely than men and non-Hispanic whites to be missing serum creatinine while older participants and non-Hispanic blacks were more likely than younger individuals and non-Hispanic whites to be missing urinary albumin or urinary creatinine data.

Of relevance to the current analysis, variables collected during the in-home interview were age, race-ethnicity, gender, cigarette smoking, a previous diagnosis of diabetes mellitus, cardiovascular disease (myocardial infarction or stroke) and pharmacologic treatment for hypertension, high cholesterol, or diabetes mellitus. Self-report of a prior diagnosis of diabetes mellitus with current use of an oral hypoglycemic agent or insulin was used to define diagnosed diabetes mellitus. Body mass index (BMI) was calculated as weight in kilograms divided by height in meters squared; obesity was defined as BMI ≥30 kg/m2. Three blood pressure measurements were obtained using a standard protocol and hypertension was defined as an average systolic or diastolic blood pressure ≥140 mmHg or 90 mmHg, respectively, or current use of blood pressure lowering medication.[8]

Total cholesterol was measured with the Hitachi 704 Analyzer; high cholesterol was defined as total cholesterol ≥240 mg/dL or pharmacologic lipid lowering treatment. A sample of NHANES 1999–2004 participants were assigned to attend a morning study visit to the mobile examination center after an overnight fast. For participants who attended the morning study visit, glucose was measured on previously frozen plasma. Among participants without diagnosed diabetes mellitus who attended a morning study visit and fasted 8 hours or longer (N = 5597 for the stage 3 or 4 CKD analyses and N = 5507 for the albuminuria analyses), undiagnosed diabetes mellitus was defined as plasma glucose ≥126 mg/dL.

Serum creatinine was measured using the modified kinetic method of Jaffe (Hitachi 917 analyzer). Serum creatinine concentrations were calibrated to the assays used for the development of the Modification of Diet in Renal Disease (MDRD) study equation.[9] The simplified MDRD study equation was used to calculate eGFR.[10] Individuals with an eGFR of 15 to 59 ml/min/1.73 m2 were considered to have stage 3 or 4 CKD.[11] Urinary albumin was measured using a solid-phase fluorescence immunoassay; urinary creatinine was measured using modified kinetic method of Jaffe (Beckman Coulter Synchron AS/Astra Analyzer). Albuminuria was defined as a urinary albumin to urinary creatinine ratio = 30 mg/g.[11] The results were markedly consistent when gender specific cut-points (urinary albumin to urinary creatinine ratio = 30 mg/g for women and = 20 mg/g for men) were used to define albuminuria and therefore are not presented.[12]

The protocols for NHANES 1999–2004 were approved by the National Center for Health Statistics of the Centers for Disease Control and Prevention Institutional Review Board. All participants provided written informed consent.

Statistical Methods

The prevalence of stage 3 or 4 CKD and albuminuria, separately, was calculated by age grouping (20 to 49 years, 50 to 69 years, and ≥70 years) for the overall population and for individuals with and without a history of hypertension and diabetes mellitus, separately. Characteristics of the population were calculated, stratified by age grouping for participants with and without stage 3 or 4 CKD and with and without albuminuria, separately. The statistical significance of differences in the means and prevalence estimates for participants with and without stage 3 or 4 CKD, and with and without albuminuria, was determined using linear and logistic regression models, respectively, after adjustment for age, modeled as a continuous variable. Prevalence ratios of stage 3 or 4 CKD and albuminuria, separately, associated with non-Hispanic black race-ethnicity, female gender, cigarette smoking, obesity, hypertension, high cholesterol, diagnosed and undiagnosed diabetes mellitus, and a history of cardiovascular disease were calculated for each age grouping, separately, using log-binomial regression models. All regression models included age, race-ethnicity, gender, hypertension and diagnosed diabetes mellitus. Undiagnosed diabetes mellitus was not included in all regression models as fasting plasma glucose was measured only on a sub-sample of NHANES participants. The statistical significance of the linear trends across age group was assessed by including an interaction term (e.g., age group*hypertension) in the regression model. For these models, each person within an age grouping was assigned the median age for that grouping (i.e., 36, 57, and 76 years for participants 20 to 49, 50 to 69, and ≥70 years, respectively).

Sample weights that account for the complex survey design of NHANES including unequal probabilities of selection, over-sampling, and non-response were applied for all analyses using SUDAAN (Version 9.1; Research Triangle Institute, Research Triangle Park, NC). Standard errors were estimated using the Taylor series linearization method.