Evaluation of analytical variability of dipstick in identification of proteinuria

Evaluation of proteinuria is essential in dogs with chronic kidney disease and dipstick may give a first rapid and inexpensive estimate. In veterinary medicine no information is available about analytical variability of dipstick test for proteinuria. The aim of this study was to evaluate precision, inter-observer variability and variability between two commercially available dipsticks. Samples (74 urine supernatants stored at -20°C during a 12-month period) were collected from dogs irrespective of underlying disease. All samples were gently thawed overnight and were assayed for proteinuria with two dipsticks applying urine to the pad by the mean of a dispensable pipette. Two observers evaluated each result. The dipstick evaluation was blinded to the other observer and to the results of the other dipstick. Intra and inter-assay variability of each dipstick were performed by testing 5 samples corresponding to the 5 different semi-quantitative results of the two dipstick for 10 consecutive times and 5 consecutive days, respectively. Variability between observer and between dipsticks was evaluated with Cohen's k test. Analysis was performed for the whole set of results and grouping results as ≤ 1+ and > 2+. Precision tests showed errors for both operators and both dipsticks. Intra-assay repeatability was good (maximum 3 errors per repeatability), whereas inter-assay variability was higher, ranging from 20% (1/5) to 80% (4/5) of erroneous results, in two cases. When present, errors were found for samples with results between 1+ and 3+. Concordance between operators was good (k = 0.68 and 0.79 for dipstick 1 and dipstick 2, respectively) and improved grouping results in the two classes (k = 0.87 and k = 0.94). Discordance was more frequent at lower results (negative, 1+ and 2+) and one observer overestimated results with both dipstick. Concordance between dipsticks was also good (k = 0.66 and 0.74 for the two operators), rising to 0.83 and 0.90 when results were grouped. Again, errors were more frequent at lower results. Moreover, one dipstick overestimated results for both operator. Errors and discordant results were never higher than one unit in the semi-quantitative scale for both precision and concordance tests. As any other methods, analytical variability in dipstick evaluation of proteinuria exists. Subjective interpretation of the pad and, to a lesser extent, different responses of different pads could affect sample. Further studies are warranted in order to evaluate the effect of this variability on clinical decision, especially when dipstick is interpreted in association with result of the urinary protein to creatinine ratio or of urine-specific gravity.