Can an algorithm based on dipstick urine protein and urine specific gravity accurately predict proteinuria?

Analytical verification of 12 most commonly used urine dipsticks in Croatia: comparability, repeatability and accuracy

Introduction

Variability among manufacturers of urine dipsticks, respective to their accuracy and measurement range, may lead to diagnostic errors and thus create a serious risk for the patient. Our aims were to determine the level of agreement between 12 most commonly used urine dipsticks in Croatia, examine their accuracy for glucose and total protein and to test their repeatability.

Materials and methods

A total of 75 urine samples were used to examine comparability and accuracy of 12 dipstick brands (Combur 10 TestM, ChoiceLine 10, Combur 10 TestUX, ComboStik 10M, ComboStik 11M, CombiScreen 11SYS, CombiScreen 10SL, Combina 13, Combina 11S, Combina 10M, UriGnost 11, Multistix 10SG). Agreement between each dipstick and the reference (Combur 10 TestM) was expressed as kappa coefficient (acceptable κ ≥ 0.80). Accuracy for glucose and total protein was tested by comparison with quantitative measurements on analysers: AU400 (Beckman Coulter, USA), Cobas 6000 c501 (Roche Diagnostics, Germany) and Architect plus c4000 (Abbott, USA). Repeatability was assessed on 20 replicates (acceptable > 90%).

Results

Best agreement was achieved for glucose, total protein and nitrite (11/11, k > 0.80) and the lowest for bilirubin (5/5, k < 0.60). Sensitivities for total protein were 41-75% (AU400) and 56-92% (Cobas and Architect); while specificities were 41-75% (AU400, Cobas, Architect). Dipsticks' sensitivity and specificity for glucose were 68-98%. Most of the dipsticks showed unacceptable repeatability (6/12, < 90%) for one parameter, most prominently for pH (3/12, < 90%).

Conclusions

Most commonly used dipsticks in Croatia showed low level of agreement between each other. Moreover, their repeatability varies among manufacturers and their accuracy for glucose and proteins is poor.