- Top of page
Our laboratory responded to a complaint from one of our hospital's veterinarians that our urine specific gravity (USG) results seemed too low. We had recently begun using a digital Atago PAL-USG Cat (Atago) refractometer (Atago Co., LTD Tokyo, Japan). One advantage of the Atago is that it reports USG results over a wide range (1.000–1.080), while the previous refractometer, a Schmidt and Haensch (S+H) “Goldberg type” refractometer (Schmidt and Haensch GmbH & Co, Berlin Germany), reports results only up to 1.040. Therefore, for urine samples with a specific gravity > 1.040, the urine had to be diluted with an equal amount of water to determine an exact result. To avoid the extra step of dilution, the Atago was being used for analysis of all urine samples in the hospital.
The performance of the Atago refractometer was therefore compared with the S+H refractometer. Both devices were temperature compensated. A test with distilled water consistently reported a specific gravity (SG) of 1.000 at room temperature by both refractometers, as expected and recommended. Urine samples submitted for routine diagnosis from hospitalized dogs and cats were collected over a period of 2 weeks. The signalment of the animals was not recorded. Samples were analyzed immediately by both instruments without dilution. Samples were always at room temperature. Most or all samples were voided. Initial samples with S+H results > 1.040 were excluded. Results from both instruments were initially available from 35 samples. Additional samples (paired results from 9 dogs and 3 cats) were collected over a period of one week, which had USG > 1.040 by the S+H. These were diluted manually by pipette with an equal amount of water to determine the exact USG on the S+H. To determine the accuracy of SG determinations by each refractometer, serial dilutions of 10% glucose (10 ml ampule with 300 mg/ml glucose, Fresenius Kabi Sweden, Uppsala Sweden), 10% NaCl and 3% albumin in water were determined with both instruments and compared with expected values. Solutions were made from chemical grade albumin and NaCl (Sigma-Aldrich Sweden AB, Stockholm, Sweden), and a 10% glucose solution for intravenous injection was diluted with distilled water. Statistical analysis was preformed with Analyze-it software v 2.21 (Analyze-it, Leeds, UK).
In total, 47 paired results of USG from both refractometers were available for comparison. The refractive index in refractometry consists of 2 components, water and the solutes. While water has a constant refractive index equal to one, any further refractive index changes are due to the concentrations of solutes. Therefore, the calculations were based on USG -1 to reflect only the variable effect of solutes. Spearman correlation was excellent (r = .99), and in the Passing–Bablok identity plot, paired results appeared linear (Figure 1A), but there was prominent proportional negative error for the Atago results in the Bland–Altman difference plot (Figure 1B). The linear regression had a slope of 0.74, with a 95% confidence interval (CI) of 0.71–0.76. The CI does not include 1.00, which indicates a proportional error. Intercept was 0.001 with a 95% CI of −0.002–0.003, which does include 0.00.
Figure 1. Comparison of specific gravity results determined by a Schmidt and Haensch (S+H) and an Atago PAL-USG (Atago) refractometer in urine samples from 47 dogs and cats. The values are given as urine specific gravity (USG) – 1 to reflect the variable effect of solute on the USG and not the constant water effect (1.0). (A) Passing–Bablok plot. The solid blue line represents the fitting line, and the solid gray line is the identity line. Dotted lines along the solid blue line indicate the 95% confidence intervals. (B) Bland–Altman difference plot. The gray line at 0.000 is the identity line. The dark blue, solid line indicates a bias of −0.008. The dashed blue-green lines represent the 95% limits of agreement of −0.0197–0.003. The dotted lines indicate the 95% confidence intervals about the limits of agreement and bias. There is a negative proportional error of the Atago results compared to the S+H results.
Download figure to PowerPoint
Overall, the S+H determined higher SG for all solutions in water. Also, the accuracy for serially diluted 10% glucose was higher for the S+H refractometer readouts, which basically agreed with the expected values (Table 1), while the Atago results were consistently lower. In contrast, the results from neither instrument matched expected results for various concentrations of NaCl, although the results from the S+H refractometer were closer to the expected numbers than the Atago. Likewise, the various albumin solutions were not accurately measured by either refractometer; however, the results from the Atago refractometer were closer to expected values. Note that the SG of a 10% glucose solution is expected to be 1.040, while the SG of a 10% NaCl solution is expected to be 1.060–1.070, and urine with a specific gravity of 1.040 is twice as concentrated as 1.020. The specific gravity of 1% albumin is expected to be approximately 1.025. Note that commercial protein standards should not be used for calibration of refractometers because they do not give accurate results due to the presence of other substances in the solution.
Table 1. Specific gravity results determined by a Schmidt and Haensch (S+H) and an Atago PAL-USG Cat (Atago) refractometer in serial dilutions of solutions of 10% glucose, 10% NaCl and 3% albumin, and published values.
A SG threshold of > 1.030 is commonly used clinically to indicate adequate renal function in dogs. In 10 samples, the S+H results were consistently > 1.030, while Atago readouts of the same samples were < 1.030 (range 1.023–1.028), which could have potentially resulted in different clinical conclusions being reached for these 10 animals. This example illustrates how the use of different refractometers may affect clinical interpretation.
Calibration of refractometers to 1.000 with distilled water does not ensure adequate refractometer performance based on our results. Both instruments reported 1.000 with water, but yielded very different results with increasing concentrations of urine from clinical cases and with standard solutions of glucose, sodium, and albumin. Calibration of a refractometer to a solute concentration in the clinically critical measurement range (eg, 1.030) should correct proportional errors in the slope better than calibrating only at a zero value with water. A solution of NaCl has been suggested for a control reagent for USG, where a solution of 3.5% salt should give a SG of 1.026. However, in our study, neither of the 2 tested instruments yielded expected results for different salt solutions. Clearly, further studies are required before recommending a saline control for clinical refractometer quality control.
Solutes usually present in urine affecting USG include urea, creatinine, phosphate, and sulfate. Therefore, a solution of those solutes may be more suitable than NaCl, albumin, or glucose control solutions. Albumin and glucose concentrations in normal urine are too low to affect SG, while protein and glucose in urine are not primarily detected by SG analysis, but by more specific tests for glucose or albumin.[3, 5]
Refractometers report a calculated SG based on refractive index of urine.[1, 6] Feline urine has been reported to have a lower SG than human or canine urine at a given refractive index.[1, 7, 6] A recommended conversion calculation in feline specific gravity is (0.846 × medical refractometer specific gravity) + 0.154. Presumably, both refractometers measured the refractive index correctly, but reported different SG results. This would explain the excellent correlation of results in the presence of a prominent proportional error. Currently, the specific feline conversion scale for feline urine is based on a report from 1956, although no conclusion was made on a particular conversion factor to be used.[1, 7] In brief, total solids in 233 urine samples were compared with the specific gravity or “instrumental increment of refractometer” in those urines. The 22 feline urine samples usually had higher SG results than 190 human urine samples and 21 canine urine samples, but feline results were on the same identity line as human and canine results in their Figures 2 and 3.
In this study, no differences were seen between feline and canine urine. It is unclear why feline USG calculation from refractive index should require a different conversion factor. The S+H refractometer is a standard medical refractometer, which performed better with various solutions of known concentrations of glucose and salt than the Atago refractometer designed especially for feline urine. Clearly, different refractometers reported different SG results either from split urine samples or from serially standard solutions of NaCl, glucose and albumin, and the former observation is of concern for clinical diagnoses.
A final assessment on the level of performance of either 2 tested refractometers for either canine or feline urine is not possible, as the number of specified samples was limited and the total solids in the urines examined were not determined by a gold standard method. Limitations of our study also include potential dilution errors of the various solutions including urines. Finally, the reading of the S+H refractometer results is subjective and variation among observers is possible.
Disclosure: The authors have indicated that they have no affiliations or financial involvement with any organization or entity with a financial interest in, or in financial competition with, the subject matter or materials discussed in this article.