Components of biological variation of some serum analytes in hospitalized pregnant women.

ported elsewhere. Burkhardt et al. (4) collected 791 urine specimens and compared the results obtained by refractometer, urinometer, and reagent stick. They reported that 90% of the stick measurements were within 0.005 units of the corresponding urinometer results; correction for pH modestly improved the results. Zack (5) reported a similar degree of correlation. In both of these studies the authors used linear regression analysis, an approach that can be criticized because the sticks report a discrete result. However, we found that non-parametric testing yields similar relationships. Burkhardt et al. concluded that the reagent stick would prove to have substantial clinical utility. Zack, on the other hand, from largely similar data, reached the opposite conclusion. In our opinion, the correct viewpoint is a function of the accuracy required. For most clinical purposes, such as interpreting the urinary sediment, or obtaining routine insight into kidney function (1), the reagent stick would be sufficient. However, for evaluating hyponatremic disorders (e.g., the syndrome of inappropriate anti-diuretic hormone, or the response to exogenous anti-diuretic hormone), or for determining the maximal diluting capacity of the kidney, a more sensitive test, such as the measurement of urine osmolality, is required. Clinicians familiar with these disorders will order the more-sensitive tests when needed. In addition, the sticks provide the opportunity for home monitoring of urine specific gravity. In certain kidney stone conditions, diluting the urine is an important aspect of management (2), but patients generally have no way of determining compliance to such regimens. In these instances, the reagent sticks could be a useful and inexpensive aid.