We assessed the value of 1) a spot urine test for diagnosing hypokalemia caused by renal potassium loss, and 2) actual 24-hour urine potassium excretion (24 hUK-actual) for diagnosing hypokalemia caused by renal potassium loss in patients treated with potassium supplementation. The study population was from Southwest China.

Patients with hypokalemia were divided into 2 groups: hypokalemia caused by renal potassium loss (n = 67) and hypokalemia caused by extrarenal potassium loss (n = 63). Urine potassium concentration (UK), urine creatinine concentration (UCr), urine potassium-creatinine ratio (UK/UCr), fractional excretion of potassium (FEK), transtubular potassium concentration gradient (TTKG), and 24-h urine potassium excretion (24 hUK-calculated) were derived from spot urine samples collected on admission, before initiation of therapy. Patients received intravenous potassium chloride 0.4 or 0.6 g/h. 24 hUK-actual was detected in patients whose serum potassium did not return to normal after 24 h of therapy.

Patients with hypokalemia caused by renal potassium loss had significantly higher UK, UK/UCr, FEK, TTKG and 24 hUK-calculated compared to patients with hypokalemia caused by extrarenal potassium loss (P < 0.05). FEK predicted renal potassium loss in hypokalemia with high accuracy at a cut-off of 9.29% (sensitivity, 80.6%; specificity, 85.7%). The area under the curve for 24 hUK-actual in predicting renal potassium loss in patients with hypokalemia treated with low or high-dose potassium chloride infusion were 0.939 or 0.956, respectively. On the spot urine test, FEK showed the highest correlation with 24 hUK-actual during low or high dose potassium chloride infusion (r = 0.831, p < 0.001 or r = 0.764, p < 0.001).

FEK from a spot urine sample represents a convenient and reliable parameter to predict renal potassium loss in patients with hypokalemia.