Current issues of ACP Journal Club are published in Annals of Internal Medicine


Prediction of siabetic nephropathy from untimed urine specimens

ACP J Club. 1992 Jan-Feb;116:19. doi:10.7326/ACPJC-1992-116-1-019

Source Citation

Nelson RG, Knowler WC, Pettitt DJ, et al. Assessment of risk of overt nephropathy in diabetic patients from albumin excretion in untimed urine specimens. Arch Intern Med. 1991 Sep;151:1761-5.



To evaluate the albumin-to-creatinine ratio in single, untimed, urine specimens as a predictor of the development of overt nephropathy in diabetic Pima Indians.


Cohort study of 8 years duration.


Gila River Indian Community in Arizona.


439 diabetic Pima and Tohono O'odhan Indians (134 men, 305 women). Participants ranged in age from 25 to 84 years and had no overt nephropathy at baseline. Mean follow-up period was 4.2 years.

Description of test and diagnostic standard

Participants were examined at about 2-year intervals. They were asked to void at the beginning of the examination and approximately 2 hours later, at the end of a glucose-tolerance test. Specimens were frozen and assayed within 30 days for albumin (mg/L) and creatinine (g/L). These measurements were expressed in a ratio of mg of albumin per g of creatinine (AC). The cut point for an elevated AC ratio was ≥ 30 mg/g.

The diagnostic standard was the development of overt nephropathy, defined as the first occurrence (after the baseline measurement) of a protein-to-creatinine (PC) ratio of ≥ 1.0 g/g, equivalent to a protein excretion rate of approximately 1 g/d. Specimens with a trace of protein on dipstick measurement were assayed for total protein, and a PC ratio was calculated.

Main results

At baseline 140 persons had AC ratios of ≥ 30 mg/g, and 299 participants had AC ratios < 30 mg/g. Overt nephropathy developed in 47 (34%) of those with AC ratios above the cut point compared with 12 persons (4%) with low AC ratios {relative risk, 8.4; 95% CI, 4.6 to 15.3}. Even after controlling for age, sex, duration of diabetes, blood pressure, and 2-hour postload plasma-glucose concentration, a high AC ratio predicted overt nephropathy by multivariate analysis (P < 0.001). Other independent predictors of overt nephropathy were duration of diabetes (P < 0.001) and 2-hour postload plasma-glucose concentration (P = 0.003). The predictive value of the AC ratio was less when the duration of diabetes was longer.


An albumin-to-creatinine ratio of ≥ 30 mg/g excreted in untimed urine specimens predicted the development of overt nephropathy during an 8-year period in diabetic Pima Indians.

Source of funding: National Institute of Diabetes and Digestive and Kidney Diseases.

Article reprints not available.


Diabetes is the single largest cause of end-stage renal disease in adults, and its incidence continues to rise while the incidence of other types of end-stage renal disease has leveled off. Patients with noninsulin-dependent diabetes comprise 50% to 60% of the cases of diabetic end-stage renal disease; in some minority groups this percentage may be as high as 85% (blacks) (1, 2). Unfortunately, it has been difficult to predict who will progress to end-stage renal disease. Low levels of albuminuria have been predictive of progression to overt nephropathy both in patients with insulin-dependent and those with noninsulin-dependent diabetes.

Most of these previous studies used timed urine specimens, either overnight or 2-hour. This study addresses the predictive value of an albumin-to-creatinine ratio for a random urine specimen for subsequent development of overt nephropathy. The advantage of random urine specimens is obvious. The incidence ratios listed are impressive. If one arranges the data in a traditional 2 × 2 table, the sensitivity is 0.80, the specificity is 0.76, and the likelihood ratio for a positive test is 3.26. The positive predictive value is 0.34 and negative predictive value is 0.96 for this population. With repeat testing, these performance characteristics should improve even further.

Given the additional knowledge from other studies that tight control of blood pressure and possibly adherence to a low-protein diet can slow the decline of renal function, the ability to identify persons at risk for renal disease is imperative. The clinician can use this information to target the patients in need of more intense intervention.

Jacqueline Pugh, MD
Audie L. Murphy Memorial Veterans Hospital San Antonio, Texas