Accurate and consistent results in monitoring urine pH are essential for proper diagnosis and treatment of many diseases. For example, urine pH is important in interpretation of findings in urine sediment. Urine pH may also influence the risk of urolith formation in susceptible animals and reflect systemic acid-base abnormalities. Commercially available urine reagent strips are commonly used to determine urine pH in veterinary patients because of their convenience and low cost. However, urine reagent strips may not be sufficiently accurate for evaluation of some diseases.1,2 Portable pH meters have been advocated as an accurate alternative to reagent strips and a relatively inexpensive alternative to benchtop pH meters. Benchtop pH meters contain special electrodes that allow hydrogen ions in a solution to move through a selective barrier, eliciting a measurable voltage proportional to the solution's pH.3 To confirm these findings, the present study was designed to compare the reproducibility and accuracy of 4 portable pH meters in measuring urine pH for use in veterinary medicine. Reproducibility is defined as the ability to obtain the same value for repeated measurements of the same analyte in a sample4 and is synonymous with measurement precision. Poor reproducibility degrades the precision of a single measurement by impairing the observer's ability to discriminate change. Accuracy is defined as the ability to obtain the established or true value of an analyte in a sample and verifies the correctness of a test result.4
Materials and Methods
REPRODUCIBILITY
Urine collection and aliquot preparation—The kidneys are capable of adjusting the pH of urine in the approximate range of 5.0 to 8.0.5,6 With the goal of obtaining urine samples with pH values in this biological range, the pH of 15 voided samples collected from 15 hospitalized dogs at the University of Minnesota Veterinary Teaching Hospital was measured by use of a benchtop pH meter. Samples of similar pH were combined into four 250-mL pooled samples. The pH of 2 pooled samples was 6.11 and 6.60, hydrochloric acid was added to 1 pooled sample to yield a pH of 5.50, and sodium hydroxide was added to another pooled sample to yield a pH of 7.60. The pooled samples were refrigerated for approximately 17 hours. A micropipette was used to create seventy 2-mL aliquots from each of the 4 pooled urine samples. Aliquots were placed in capped plastic vials until the time of measurement.
Sample masking and randomization—A technician not involved in pH measurement labeled each 2-mL urine aliquot with a number from a randomly generated numeric code. Ten aliquots from each of the 4 samples were distributed to each device. After all pH measurements were completed, each 2-mL aliquot was matched to its original sample according to the random code.
pH measurement—The reproducibility of measurements from 7 devices used to measure urine pH was compared. Devices evaluated were the pH meter 430a (benchtop reference method); Checker 1,b Oakton waterproof pH tester BNC,c Extech waterproof palm pH meter,d and Omega PHH-5012e (portable pH meters); Multistixf (urine reagent strip); and pHydrion Vivid 1-11g (pH paper). Just prior to sample measurement, each meter was calibrated according to manufacturers' directions with commercial buffer solutions of pH 4.0h and 7.0.i The calibration of each device was verified after every 10 measurements. All aliquots were assayed within 3 hours of preparation. The pH of each urine aliquot was measured and recorded to the nearest 0.01 pH unit according to the manufacturer's instructions. Values of pH determined by the reagent strip and pH paper were measured and recorded to the nearest 0.5 pH unit. To eliminate interobserver variability, 1 individual performed all pH measurements.
Statistical analysis—Intra-assay precision for each method was evaluated by obtaining 10 measurements from pooled urine samples at 4 pHs. The coefficient of variation (a measure of deviation of a variable from its mean), calculated as SD multiplied by 100 and divided by the mean, was calculated for each device at each pH.7
ACCURACY
Urine collection and aliquot preparation—During an 8-day period, 201 voided urine samples were collected from 114 hospitalized dogs. Immediately after collection, seven 2-mL aliquots were removed from each sample by use of a micropipette. Aliquots were placed in capped plastic vialsj to prevent evaporation and stored at 20° to 27°C.
Sample masking and randomization—A technician not involved in pH measurement assigned a label with a randomly generated numeric code to each 2-mL aliquot. The 7 aliquots from each sample were distributed equally among the pH measurement devices. After all pH measurements were completed, each aliquot was unmasked and matched to its original sample.
pH measurement—All 2-mL urine aliquots were measured within 4 hours of collection, and pH measurements were performed in the same manner as described for the first part of the study.
Statistical analysis—A paired t test was used to identify differences between measurements yielded by the benchtop pH meter (reference method) and those obtained with the other 6 devices.8 Agreement between each device and the benchtop pH reference method was assessed graphically by use of the CLSI bias plot.9 This test is used when 1 test method is regarded as definitive; the difference between the test methods is plotted against the reference method. The concordance correlation coefficient was calculated to obtain a numeric representation of agreement.10,11 The concordance correlation coefficient assesses agreement by measuring variation from the line of identity (y = x).
Results
Reproducibility—All 4 portable pH meters had coefficients of variation < 0.425% (Table 1). The reference method had the smallest mean coefficient of variation (0.118). The mean coefficient of variation of the urine reagent strip and pH paper was approximately 9 times as great as the mean coefficient of variation of the 4 portable pH meters.
Coefficients of variation for urine pH measurements obtained with 7 pH-determining devices at 4 pHs. Coefficients of variation at each pH were calculated from 10 measurements made on each of 4 pooled urine samples from 15 dogs.
| Device | Coefficient of variation (%) | ||||
|---|---|---|---|---|---|
| pH = 5.5 | pH = 6.1 | pH = 6.8 | pH = 7.6 | Mean | |
| Benchtop meter* | 0.094 | 0.084 | 0.185 | 0.108 | 0.118 |
| Checker portable meter | 0.425 | 0.263 | 0.411 | 0.309 | 0.352 |
| Omega portable meter | 0.155 | 0.185 | 0.210 | 0.116 | 0.166 |
| Oakton portable meter | 0.18 | 0.205 | 0.290 | 0.256 | 0.233 |
| Extech portable meter | 0.192 | 0.237 | 0.297 | 0.185 | 0.228 |
| Multistix reagent strip | 2.243 | 1.988 | 1.125 | 1.051 | 1.602 |
| pHydrion Vivid paper | 4.452 | 3.456 | 0.000 | 2.669 | 2.644 |
Reference method.
Accuracy—No significant differences between pH values measured by the benchtop pH meter and those measured by any of the 4 portable pH meters were detected (Table 2). In contrast, significant differences were detected between pH values measured by the benchtop pH meter and both the reagent strip and pH paper. The 4 portable pH meters had the highest concordance correlation coefficients, indicating a tight fit along the line of identity (Figures 1–4). In comparison, pH values measured with the reagent strip and pH paper had the lowest concordance correlation coefficients and were not associated with the same degree of agreement. Those values had greater deviation from the line of identity (Figures 5 and 6). The 4 portable meters had the smallest bias values and the least spread on the CLSI bias plots, and the reagent strip and pH paper had the highest bias values and the greatest spread on the CLSI bias plots.
Summary of statistical values indicating agreement and bias for 7 pH determining devices used to measure pH in 201 urine samples from 114 dogs.
| Device | Mean ± SD pH | 95% confidence interval | P value | Concordance correlation coefficient | Mean bias(pH units) |
|---|---|---|---|---|---|
| Benchtop meter* | 6.48 ± 0.85 | 4.81–8.15 | NA | NA | 0.118 |
| Checker portable meter | 6.47 ± 0.86 | 4.78–8.16 | 0.870 | 0.9974 | –0.014 |
| Omega portable meter | 6.50 ± 0.84 | 4.85–8.15 | 0.780 | 0.9928 | 0.024 |
| Oakton portable meter | 6.47 ± 0.84 | 4.82–8.12 | 0.700 | 0.9888 | 0.033 |
| Extech portable meter | 6.45 ± 0.90 | 4.69–8.21 | 0.760 | 0.9771 | –0.027 |
| Multistix reagent strip | 6.70 ± 0.91 | 4.92–8.48 | 0.014 | 0.9249 | 0.216 |
| pHydrion Vivid paper | 6.74 ± 0.76 | 5.25–8.23 | 0.001 | 0.8435 | 0.262 |
NA = Not applicable.
See Table 1 for remainder of key.
Scatterplot depicting the relationship between urine pH measurements obtained with the Checker portable pH meterb and those obtained with a benchtop pH metera used as the reference method (A) and bias plot depicting the differences in measurements between the 2 methods plotted against the values determined via the reference method (B). Mean bias = −0.014. Dashed lines represent mean ± 2 SD. Data represent pH values from 201 free-catch urine samples collected from 114 hospitalized dogs.
Citation: Journal of the American Veterinary Medical Association 230, 3; 10.2460/javma.230.3.364
Scatterplot depicting the relationship between urine pH measurements obtained with the Omega portable pH metere and those obtained with the benchtop pH metera (A) and bias plot depicting the differences in measurements between the 2 methods plotted against the values determined via the reference method (B). Mean bias = 0.024. See Figure 1 for key.
Citation: Journal of the American Veterinary Medical Association 230, 3; 10.2460/javma.230.3.364
Scatterplot depicting the relationship between urine pH measurements obtained with the Oakton portable pH meterc and those obtained with the benchtop pH metera (A) and bias plot depicting the differences in measurements between the 2 methods plotted against the values determined via the reference method (B). Mean bias = 0.033. See Figure 1 for key.
Citation: Journal of the American Veterinary Medical Association 230, 3; 10.2460/javma.230.3.364
Scatterplot depicting the relationship between urine pH measurements obtained with the Extech portable pH meterd and those obtained with the benchtop pH metera (A) and bias plot depicting the differences in measurements between the 2 methods plotted against the values determined via the reference method (B). Mean bias = −0.027. See Figure 1 for key.
Citation: Journal of the American Veterinary Medical Association 230, 3; 10.2460/javma.230.3.364
Scatterplot depicting the relationship between urine reagent stripf pH measurements and values obtained by use of the benchtop pH metera (A) and bias plot depicting the differences in measurements between the 2 methods plotted against the values determined via the reference method (B). Mean bias = 0.216. See Figure 1 for key.
Citation: Journal of the American Veterinary Medical Association 230, 3; 10.2460/javma.230.3.364
Scatterplot depicting the relationship between pH measurements obtained by use of pH paperg and values obtained with the benchtop pH metera (A) and bias plot depicting the differences in measurements between the 2 methods plotted against the values determined via the reference method (B). Mean bias = 0.262. See Figure 1 for key.
Citation: Journal of the American Veterinary Medical Association 230, 3; 10.2460/javma.230.3.364
Reagent strips and pH paper tended to overestimate the pH values obtained from the benchtop meter (Tables 3 and 4). The reagent strip had the largest range of values at pHs of 6 to 7.5, and the pH paper had the largest range of values at pHs of 6 to 8.
Frequency, mean, and range of urine pH values obtained with a reagent strip, compared with values obtained by use of the reference benchtop pH meter. Notice that most of the reagent strip pH measurements were higher than values obtained with the reference method.
| Reagent stripf | Reference methoda pH | ||
|---|---|---|---|
| pH | Frequency* | Mean | Range |
| 5.0 | 6 | 5.2 | 5.02–5.38 |
| 5.5 | 29 | 5.51 | 5.26–5.92 |
| 6.0 | 47 | 5.8 | 5.26–6.23 |
| 6.5 | 31 | 6.34 | 6.02–6.70 |
| 7.0 | 28 | 6.84 | 6.31–7.31 |
| 7.5 | 37 | 7.3 | 6.38–7.71 |
| 8.0 | 15 | 7.81 | 7.49–8.11 |
| 8.5 | 8 | 8.19 | 8.03–8.34 |
Frequency with which a pH was measured in 201 samples.
Frequency, mean, and range of urine pH values obtained with pH paper, compared with values obtained by use of the reference benchtop pH meter. Notice that most of the pH paper measurements were higher than values obtained with the reference method.
| pH paperg | Reference methoda pH | ||
|---|---|---|---|
| pH | Frequency* | Mean | Range |
| 5.0 | 3 | 5.16 | 5.02–5.32 |
| 5.5 | 3 | 5.42 | 5.29–5.52 |
| 6.0 | 60 | 5.61 | 5.14–7.1 |
| 6.5 | 39 | 6.14 | 5.66–6.70 |
| 7.0 | 51 | 6.84 | 6.56–7.24 |
| 7.5 | 23 | 7.26 | 6.81–7.71 |
| 8.0 | 14 | 7.78 | 6.82–8.31 |
| 8.5 | 7 | 8.18 | 8.03–8.34 |
| 9.0 | 1 | NA | 8.3 |
NA = Not applicable.
See Table 3 for remainder of key.
Discussion
Results indicated that reagent strips were associated with less reproducible and less accurate measurements than pH meters. These findings are consistent with those of previous investigators. In 1 study2 of samples from 52 dogs and 109 cats, a reagent strip pH value of 7.0 corresponded to reference pH values ranging from 6.2 to 8.2. In another study1 of 40 samples from cats, a reagent strip value of 6 corresponded to reference pH values ranging from 5.81 to 7.18. We did not observe as great a magnitude of difference between reagent strip and pH meter measurements in the present study. In this study, values generated by reagent strips were ± 0.37 pH units from the reference value generated by the benchtop pH meter, a difference that was within the manufacturers' reported range of ± 0.5 units.
The pH values obtained with all 4 portable pH meters were reproducible and accurate. All coefficients of variation were < 0.5%, all concordance correlation coefficients were > 0.97, and all bias values were within 0.1 pH unit. No significant difference was detected between pH values obtained with the 4 portable pH meters, although the Checker 1 meter had the highest concordance correlation coefficient and the smallest bias value.
Reproducible and accurate urine pH measurements are important in diagnosis and treatment of several diseases. For example, urine pH is an important determinant of struvite crystal formation, and changing urine pH has a proportionately greater effect on the struvite activity product than changing the concentration of 1 or more of the components of struvite.12 Decreasing urine pH promotes formation of urine that is undersaturated with struvite. In that environment, struvite crystal formation and growth will not occur and preformed struvite uroliths may dissolve.13 Cats with urine pH values < 6.4 are unlikely to form struvite uroliths.14 Acidification of the urine may not be appropriate, however, in the management of other types of uroliths. Epidemiologic data in dogs and cats suggest that acidifying diets, especially those that result in urine pH < 6.29, may increase the risk of calcium oxalate formation.15,16
Urine pH is also an important factor in managing urinary tract infections with certain anti-infective drugs. For example, urine pH values < 6.1 promote conversion of methenamine to bacteriostatic concentrations of formaldehyde.17 Knowledge of urine pH can also be used to confirm diagnosis of renal tubular acidosis. Proximal renal tubular acidosis is associated with urine pH values > 6 after bicarbonate administration,18 and distal renal tubular acidosis is associated with impaired ability to acidify urine below a pH of 5.5 after ammonium chloride administration.19
Results indicated that urine reagent strips may be used to obtain estimates of pH for routine urinalysis but should not be relied upon when consistent and accurate pH measurements are crucial for diagnosis, prevention, and management of disease. In those situations, urine pH should be determined by use of a portable or benchtop pH meter.
ABBREVIATIONS
| CLSI | Clinical and Laboratory Standards Institute |
pH meter 430, Corning, Corning, NY.
Checker 1, Hanna Instruments, Woonsocket, RI.
Oakton waterproof pH tester BNC, Oakton Instruments, Vernon Hills, Ill.
Extech waterproof palm pH meter, model PH 220, Extech Instruments, Waltham, Mass.
Omega PHH-5012, Omega Engineering, Stamford, Conn.
Multistix, Bayer Corp, Elkhart, Ind.
pHydrion Vivid 1-11, Micro Essential Laboratory, Brooklyn, NY.
Buffer pH 4.0, EMD Chemicals Inc, Gibbstown, NJ.
Buffer pH 7.0, EMD Chemicals Inc, Gibbstown, NJ.
Copolymer plastic vial, 7 mL unlined, Research Products International, Mount Prospect, Ill.
References
- 1↑
Raskin RE, Murray KA, Levy JK. Comparison of home monitoring methods for feline urine pH measurement. Vet Clin Pathol 2002;31:51–55.
- 2↑
Heuter KJ, Buffington CA, Chew DJ. Agreement between two methods for measuring urine pH in cats and dogs. J Am Vet Med Assoc 1998;213:996–998.
- 3↑
Sensorex Corporation Web site. pH Electrode Technical Education, 2006. Available at: www.sensorex.com/support/education/pH_education.html. Accessed Aug 7, 2006.
- 4↑
Osborne CA, Stevens JB. “Prophet”-ing more from urinalysis: maximizing reproducible test results. In: Osborne CA, Stevens JB, eds.Urinalysis: guide to compassionate patient care. Shawnee Mission, Kan: Bayer Co, 1999;51–63.
- 5
Osborne CA, Stevens JB. Biochemical analysis of urine: indications, methods, interpretation. In: Osborne CA, Stevens JB, eds.Urinalysis: guide to compassionate patient care. Shawnee Mission, Kan: Bayer Co, 1999;88–92.
- 6
Ryan JP, Irwin JA. Practical studies on urine demonstrating principles of clinical and veterinary significance. Biochem Mol Biol Educ 2002;30:98–100.
- 7↑
Motlusky H. The standard deviation. In: Moltusky H, ed. Intuitive biostatistics. New York: Oxford University Press, 1995;22–30.
- 8↑
Motlusky H. Comparing two paired groups: paired t and wilcoxon tests. In: Moltusky H, ed. Intuitive biostatistics. New York: Oxford University Press, 1995;225–229.
- 9↑
National Committee for Clinical Laboratory Standards. Method comparison and bias estimation using patient samples, approved guideline. NCCLS EP9-A. Wayne, Pa: NCCLS, 1995.
- 10
St Laurent RT. Evaluating agreement with a gold standard method in comparison studies. Biometrics 1998;54:537–545.
- 11
Lin LI-K. A concordance correlation coefficient to evaluate reproducibility. Biometrics 1989;45:255–268.
- 12↑
Markwell PJ, Buffington CA, Smith BHE. The effect of diet on lower urinary tract diseases in cats. J Nutr 1998;218:2735–2757.
- 13↑
Osborne CA, Lulich JP, Kruger JM. Medical dissolution of feline struvite urocystoliths. J Am Vet Med Assoc 1990;196:1053–1063.
- 14↑
Buffington CA, Rogers QR, Morris JG. Effect of diet on struvite activity product in feline urine. Am J Vet Res 1990;51:2025–2029.
- 15
Kirk CA, Ling GV, Franti CE, et al. Evaluation of factors associated with development of calcium oxalate urolithiasis in cats. J Am Vet Med Assoc 1995;207:1429–1434.
- 16
Lekcharoensuk C, Osborne CA, Lulich JP, et al. Association between dietary factors and calcium oxalate and magnesium ammonium phosphate urolithiasis in cats. J Am Vet Med Assoc 2001;219:1228–1237.
- 17↑
Strom JG Jr, Jun HW. Effect of urine pH and ascorbic acid on the rate of conversion of methenamine to formaldehyde. Biopharm Drug Dispos 1993;14:61–69.
- 18↑
Igarashi T, Sekine T, Watanabe H. Molecular basis of proximal renal tubular acidosis. J Nephrol 2002;15 (suppl 5):S135–S141.
- 19↑
Unwin RJ, Shirley DG, Capasso G. Urinary acidification and distal renal tubular acidosis. J Nephrol 2002;15 (suppl 5):S142–S150.
