The latter three were however only statistically significant in either the analysis with UACR as a continuous variable or in the analysis with UACR quartiles and should be interpreted with caution. On death caused by circulatory and endocrine, nutritional and metabolic diseases, the LY2157299 observed positive and linear relationships with UACR levels are in line with previous studies. Likewise, our results of UACR and all-cause mortality are in line with previous studies. This association has been extensively studied in diabetics in particular –both micro- and normo-albuminurics. A collaborative meta-analysis of prospective general population studies found ACR 1.1 mg/mmol or more to be an independent predictor of mortality risk in the general population. Our study is an extension of the previous studies since it also examines other specific causes of death than cardiovascular disease. Regarding death caused by respiratory disease, our results are in line with Bulcun et al who found that UACR was significantly higher in patients with COPD than in controls. Similarly, Casanova et al found microalbuminuria to be common in COPD patients and associated with hypoxemia independent of other cardiovascular risk factors. The progressive airway limitation and destruction of pulmonary capillaries in COPD lead to the characteristic ventilation/perfusion abnormality which in turn causes hypoxemia. Hypoxia is thought to cause endothelial dysfunction which is closely ASP1517 related to albuminuria. Albuminuria is considered a marker of small vessel disease and is associated with risk of hypertension, obesity and glucose levels. Cognitive decline is frequently attributed to microvascular disease in the brain, and the mentioned risk factors have been shown to predict dementia later in life. The positive association between albuminuria and mental and behavioral disorders that mostly consists of dementia may therefore reflect the cumulative vascular damage over years related to hypertension, abnormal glucose metabolism, and other risk factor. Regarding the possible U-shape of the association between UACR status and all-cause and endocrine, nutritional and metabolic disease mortality, it is somewhat in line with a large study by Kovesdy et al who reported a similar U-shape in patients with advanced CKD of the associations between UACR and allcause mortality and progressive CKD. They found that very low levels of UACR were associated with a higher risk in this 
subgroup –maybe reflecting an inability to adapt to lower renal perfusion pressures in CKD– and that the optimal range in this group was 10–19 mg/g. However, this explanation does not suffice in explaining why a similar U-shape is seen in our general population study. The U-shape may reflect the higher mortality among persons underweight and patients with other comorbidities. Whether there is a safe threshold of albuminuria is still under debate because even urine albumin in the upper normal range bears a significant risk: albuminuria well below what is usually defined as microalbuminuria is a strong predictor of cardiovascular morbidity and mortality and any degree of measurable albuminuria bears significant cardiovascular risk. Likewise, a study found that a baseline urinary ACR $5 mg/g, a level not traditionally considered clinically significant, is independently associated with faster decline in cognitive function.