Of 11,656 participants who presented for the fourth study visit, 10,975 were included in the analysis because of exclusions for prevalent heart failure (n = 455), missing UACR information (n = 160), history of dialysis or transplant (n = 6), eGFR <30 mL/min/1.73 m2 (n = 30), and race other than black or white (n = 30). Compared with individuals included in the study, individuals with missing UACR values had similar characteristics of age, sex, education level, and smoking status, but were more likely to be black, have hypertension, or have diabetes. In study participants, mean ± standard deviation and median UACRs were 22.9 ± 184.9 and 3.7 mg/g, respectively. Most individuals were found to have optimal UACRs (n = 6,733; 61.4% of cohort), with 19.4% (n = 2,128) and 11.8% (n =1,294) of individuals having intermediate-normal and high-normal UACRs, respectively. Prevalences of microalbuminuria (n = 678) and macroalbuminuria (n = 142) were 6.2% and 1.3%, respectively.
lists baseline characteristics of participants by UACR category. Higher UACR was associated with older age, lower educational attainment, and higher prevalences of CHD, hypertension, and diabetes. Participants with micro- or macroalbuminuria had on average lower eGFRs and high-density lipoprotein cholesterol levels and higher systolic blood pressures and BMI than participants with normal UACR.
Baseline Characteristics of Participants by UACR Category
Participants were followed up for a median of 8.3 years and experienced 344 incident heart failure events for an incident rate of 3.9 events/1,000 person-years. Incidence rates progressively increased with higher levels of UACR (P <0.001 for trend). After adjustment for covariates, intermediate-normal UACR was associated with a 54% increase in the relative hazard of heart failure compared with normal UACR, whereas high-normal UACR was associated with a 91% increase in the relative hazard of heart failure (). shows the incidence of heart failure; survival curves were statistically different both globally (P < 0.001) and among all individual curves (P < 0.01 for all comparisons).
Incident Heart Failure Events by UACR Category
Figure 1 Incidence of heart failure by albuminuria category in 10,975 Atherosclerosis Risk in Communities (ARIC) Study participants. Albuminuria categories were based on urinary albumin-creatinine ratios (UACRs) as macroalbuminuria (UACR ≥300 mg/g), microalbuminuria (more ...)
In the analysis of UACR as a continuous log-transformed variable, the relative hazard of heart failure was 1.15 (95% confidence interval [CI], 1.10–1.21) for each doubling of UACR (eg, 20 vs 10 or 10 vs 5 mg/g). The risk associated with doubling of UACR was similar after exclusion of individuals with micro- or macroalbuminuria (HR, 1.20; 95% CI, 1.10–1.31). Overall, we found a linear relationship between level of albuminuria and relative hazard of the development of heart failure. shows the linear spline model of continuous UACR with knots at 10, 30, and 300 mg/g. The relative hazard begins to increase linearly from the baseline value of 1 mg/g.
Figure 2 Adjusted relative hazard of heart failure by continuous level of urinary albumin-creatinine ratio (UACR) in 10,975 Atherosclerosis Risk in Communities (ARIC) Study participants. Reference point is UACR of 1 mg/g. Graph represents a linear spline model, (more ...)
In the secondary analysis using the more sensitive definition of heart failure, there were 752 incident cases of heart failure over a median of 8.2 years, for an incident rate of 8.7 cases/1,000 person-years. The incidence of heart failure appeared to increase with each category of UACR (P
< 0.001 for trend; Fig S1
, available as online supplementary material
). After adjustment for covariates, the HR of heart failure was significantly increased in individuals with intermediate-normal UACR (1.25; 95% CI, 1.01–1.55), high-normal UACR (1.68; 95% CI, 1.34–2.11), microalbuminuria (2.35; 95% CI, 1.86–2.96), and macroalbuminuria (4.56; 95% CI, 3.29–6.31; Table S1
No interaction was found between UACR category and race, sex, diabetes, hypertension, or CHD (P > 0.1 for all comparisons). In subgroup analysis, continuous UACR was associated with increased risk of incident heart failure in individuals with and without hypertension, with and without diabetes, with and without baseline CHD, and free from CHD during follow up (). This relationship persisted even after excluding individuals with micro- or macroalbuminuria. In individuals with UACR <30 mg/g who were free from diabetes and hypertension and censored at the time of the CHD event, doubling of UACR was associated with a 49% increase in the relative hazard of heart failure.
Adjusted HR of Heart Failure for Doubling of UACR
No interaction was found between UACR and eGFR category (P
= 0.4). Subgroup analysis by eGFR category showed that increased levels of albuminuria were associated with increased risk of incident heart failure along categories of eGFR (). This association was particularly evident in the analysis that used the more sensitive definition of heart failure, resulting in increased power (Table S2
). In this analysis, the trend for association of albuminuria and incident heart failure was highly significant along all categories of eGFR, with the exception of eGFR >120 mL/min/1.73 m2
trend =0.2; P
trend <0.01 for all other categories).
Adjusted Relative Hazard of Heart Failure by Category of UACR and eGFR
In the sensitivity analysis using calculated adjusted UACR based on sex and race, we obtained results similar to results in our primary analyses. Adjusted relative hazards of heart failure for intermediate-normal adjusted UACR (1.57; 95% CI, 1.14–2.15), high-normal adjusted UACR (1.95; 95% CI, 1.38–2.67), microalbuminuria (2.46; 95% CI, 1.75–3.48), and macroalbuminuria (3.53; 95% CI, 2.21–5.66) were all increased compared with optimal adjusted UACR. In the analysis of adjusted UACR as a continuous log-transformed variable, the relative hazard of heart failure was 1.15 (95% CI, 1.10–1.21) for each doubling in adjusted UACR.