Is kidney size a useful predictor of renal function in the elderly?

Despite its widespread use, the diagnostic yield of ultrasonography to detect early stages of reduced renal function in the elderly remains doubtful.

Sixty-two patients (age: 74.8 +/- 6.7 years, 30 females) attending a geriatric ward were examined. Non-insulin-dependent diabetes was present in 28 patients and 32 were hypertensive. Exclusion criteria were previously known renal failure, an elevated serum creatinine above 2 mg/dl at admission or apparent cognitive impairment. Inulin clearance (C(IN)), Cockcroft estimation and urinary albumin excretion rate (UAER) were determined. Renal parenchymal volume, renal volume and mean renal length were calculated using data obtained by ultrasonography.

A multiple regression analysis identified C(IN) as the main determining factor on renal parenchymal volume (beta = 0.360; p = 0.004). Body height showed additional influence (beta = 0.295; p = 0.018), but age, gender, diabetes, UAER and hypertension did not. Cockcroft estimation, ultrasonographic markers of renal size and further confounding variables were evaluated in a subsequent multiple regression analysis. Cockcroft estimation explained the major part of the variance (beta = 0.783; p < 0.001) and the occurrence of diabetes showed a borderline additional effect. But there was no supplementary influence of any ultrasonographic parameter. In 11 patients, C(IN) was reduced to < or = 60 ml/min. ROC analysis revealed poor diagnostic efficacy of all ultrasonographic parameters (AUC < 0.7336) compared to Cockcroft estimation (AUC = 0.8718).

Although a linear relation between C(IN) and renal parenchymal volume could be shown, ultrasonography cannot replace or complement Cockcroft estimation in order to detect early reduced renal function in the elderly. Nevertheless, reduced renal parenchymal volume may occur prior to elevation of serum creatinine in the elderly and should be noted carefully on routine ultrasonographic examinations.