Abstract
OBJECTIVE To investigate the performance of a portable blood glucose meter (PBGM) designed for use in humans (hPBGM) and a PBGM designed for use in dogs and cats (vPBGM) when measuring blood glucose (BG) concentration in tigers (Panthera tigris) and lions (Panthera leo).
DESIGN Method comparison and diagnostic accuracy study.
SAMPLES 53 blood samples from tigers (n = 27) and lions (26).
PROCEDURES BG concentration was measured with 2 identical hPBGMs, 2 identical vPBGMs, and a reference laboratory analyzer. Bland-Altman bias plots and Passing-Bablok regression analysis were used to assess agreement. Sensitivity, specificity, and positive and negative predictive values with corresponding 95% confidence intervals were calculated for use in assessing diagnostic accuracy of the investigated PBGMs.
RESULTS Bias (95% limits of agreement) was −4.3 mg/dL (−46.3 to 37.6 mg/dL) for the hPBGM, −9.3 mg/dL (−64.6 to 46.0 mg/dL) for the vPBGM on canine setting, and 2.3 mg/dL (−47.9 to 52.6 mg/dL) for the vPBGM on feline setting. The hPBGM had better overall repeatability (coefficient of variation, 3.73%) than the vPBGM on canine (9.29%) or feline (9.44%) setting. Total error for the hPBGM, vPBGM on canine setting, and vPBGM on feline setting was 11.8%, 27.7%, and 20.9%, respectively. None of the PBGMs complied with the maximum allowable total error suggested by current guidelines when measuring BG in tigers and lions with hypo-, normo-, or hyperglycemia.
CONCLUSIONS AND CLINICAL RELEVANCE Results indicated that the PBGMs evaluated were inadequate for measuring BG concentration in tigers and lions.
Glucose homeostasis has been increasingly recognized as a priority in medically managing critically ill humans and animals,1–3 and BG concentration measurements may assist in diagnosing diseases in veterinary patients.4–7 Further, BG concentration may be a prognostic indicator because hypo- and hyperglycemia are associated with increased morbidity and mortality rates in certain settings.1 To the authors’ knowledge, the first evidence that BG concentrations should be monitored during anesthesia of tigers and lions was provided in a recent study8 in which mean ± SD BG concentration of 11 female tigers (Panthera tigris) and lions (Panthera leo) increased from 143.0 ± 15.5 mg/dL at baseline to 192.0 ± 19.4 mg/dL after 1.5 hours of anesthesia. Although BG concentration was not the primary interest in that study,8 the results suggest that changes in BG concentrations may occur during anesthesia in these species.
Measuring BG concentration with POC testing tools (ie, instruments used to perform laboratory analyses in close proximity to patients and outside of conventional reference laboratory settings) provides fast results, and such POC testing tools have become invaluable in human and veterinary medicine.9,10 Several POC testing tools for measuring BG concentration are currently available and include noninstrumental systems (ie, reagent test strips), portable analyzers (ie, PBGMs), and benchtop analyzers (ie, automated biochemical analyzers).11–13 We believe that the most versatile POC tools for measuring BG concentration are PBGMs, which are basically handheld instruments that use reagent test strips and provide immediate results. These PBGMs may be useful in veterinary medicine for zoological and wild animals14 because the instruments can be transported and used almost anywhere, including at extreme altitudes,15 and may be ideal for monitoring BG concentration measurements during surgical procedures in the field. However, most commercially available PBGMs were designed for measuring BG concentrations in humans, and instrument accuracy has been a constant concern in human medicine since initial PBGM development in the 1970s.10,16 Recently, PBGMs designed for veterinary use have been developed; yet, evidence shows that, because PBGM results may differ substantially from true BG concentrations (determined on the basis of reference laboratory analyzer results), a PBGM's performance should always be evaluated in the target species before implementing clinical use of the instrument with that species.17–20
Typically, performance of PBGMs has been evaluated by comparing results obtained from the PBGM with results obtained from a reference laboratory analyzer that uses the hexokinase reference method of analysis.21,22 In veterinary medicine, agreement of PBGMs with reference laboratory analyzers has been evaluated in dogs,21 cats,22 horses,23 seabirds,17 Hispaniolan Amazon parrots,18 deer,14 sheep,19 cattle,24 alpacas,20,25 ferrets,26 and rabbits.27 Discrepancies between results from PBGMs and results from reference laboratory analyzers have been demonstrated2 and may cause clinical errors. Performance of PBGMs in some species may also be affected by blood sample characteristics. For example, when blood samples with Hct outside of reference limits are tested, PBGM accuracy is affected in dogs28 and rabbits,27 but not in cats.29
Despite PBGMs’ attractiveness on the basis of the immediate availability of results, small sample volume required, and transportability, evaluation of PBGM performance in tigers and lions is needed prior to clinical use of the instruments with these species. Therefore, the objectives of the study reported here were to investigate the performances of an hPBGM and a vPBGM when measuring BG concentration in tigers and lions, assess agreement between the PBGMs and a reference laboratory analyzer, and determine potential effects on clinical decision-making associated with the use of these PBGMs when measuring BG concentration in tigers and lions. Furthermore, we investigated potential impacts of Hct and BG concentration ranges on the performance of these PBGMs. We hypothesized that the vPBGM set to measure BG concentration in cats (feline setting) would be the most appropriate for measuring BG concentration in tigers and lions and that Hct would influence performance of the PBGMs.
Materials and Methods
Samples
All tigers and lions from various zoological institutions and animal sanctuaries in Italy undergoing blood sampling between March 20, 2016, and September 20, 2016, for diagnostic or screening reasons unrelated to the present study were eligible for inclusion. Healthy and diseased tigers and lions were included to better ensure wide ranges of BG concentration and Hct for evaluation. The study was performed in compliance with Directive 2010/63/EU on the protection of animals used for scientific purposes, and the owners or curators of the animals gave written informed consent for inclusion of blood samples obtained from their animals.
Procedures
For reasons unrelated to the present study, tigers and lions were sedated with a standard sedative protocol that included detomidine (50 μg/kg [22.7 μg/lb]), ketamine (2 mg/kg [0.9 mg/lb]), and methadone (0.1 mg/kg [0.045 mg/kg]) combined into a single syringe for IM administration by remote injection. Once an animal was sedate (approx 20 minutes after the injection), a venipuncture site over the medial saphenous vein was clipped, then swabbed with 70% alcohol. Each blood sample was collected with a 22-gauge needle attached to a 5-mL plastic syringe. The needle was removed from the syringe, and a drop of blood was immediately deposited onto a glass slide and analyzed in duplicate as previously described27 with 2 identical hPBGMsa and 2 identical vPBGMs.b Of the 2 vPBGMs, one was set to measure BG concentration in dogs (canine setting), and the other was set to measure BG concentration in cats (feline setting). To minimize procedure time, venipuncture and blood sample processing were performed by the same operator (MC). The order in which the PBGMs were used was randomized with a random sequence generator to avoid bias caused by the effect of time31 and blood sample drop size.21 Blood was applied on only 1 side of the vPBGM test strips by touching a corner of a test strip to a drop of blood. Blood was not scraped onto the test strips. The remaining blood in the syringe, minus that needed for Hct determination, was immediately transferred into glass tubes without any additives. For each blood sample, Hct was determined in duplicate by filling a pair of 32 × 0.8-mm heparinized glass capillary tubes that, 30 minutes after blood collection, were centrifuged in the field with a portable centrifuge at 1,300 × g for 10 minutes. To further minimize potential delay in procedures, the PBGMs and portable centrifuge were located immediately adjacent to each other at the field site where blood collection occurred. In addition, serum from each blood sample was analyzed with a reference laboratory analyzerc within 8 hours after collection to determine the reference BG concentration. Reference laboratory personnel were not aware of results obtained with the PBGMs, and the PBGM operator was not aware of results obtained with the reference laboratory analyzer until the end of data collection.
Equipment and quality control
The PBGMs and test strips were maintained and operated within 20° to 24°C (68° to 75.2°F) and away from sunlight and heat. The test strips were stored in their original vials, which were kept closed except for the brief time required to remove test strips as needed for each test. To account for lot-to-lot variability, 2 lots of test strips for each PBGM were used.32
hPBGM—The 2 identical hPBGMs involved use of single-use test stripsd and determined BG concentration amperometrically by means of a pyrroloquinoline quinone-glucose dehydrogenase-catalyzed reaction. Results were obtained in approximately 5 seconds, and the manufacturer's reported range for BG concentration measurements was 10 to 600 mg/dL.33
vPBGM—The 2 identical vPBGMs involved use of single-use test stripse and determined BG concentration colorimetrically by means of a flavin-adenine dinucleotide-glucose dehydrogenase-catalyzed reaction.34 The vPBGMs provided results in approximately 8 seconds, and the manufacturer's reported range for BG concentration measurements was 20 to 750 mg/dL.34 In addition to recording BG concentration results obtained with one of the vPBGMs on canine setting and the other on feline setting, each of these results was then also converted, with the manufacturer's proprietary algorithmf that had been used in a previous study,27 to a corresponding concentration for the nonselected setting so that 2 results for each species setting were recorded for each measurement.
A reference laboratory analyzer that measured BG concentration by use of an enzymatic hexokinase oxidase reaction was used as the criterion standard. Linear calibration of the laboratory analyzer was performed weekly during the study period.
Statistical analysis
Agreement of the PBGMs with the reference laboratory analyzer in measuring BG concentration in tigers and lions was assessed with Bland-Altman bias plots and Passing-Bablok regression analysis. The 95% LoA were determined from the Bland-Altman plots35 as bias ± 1.96 SD, with bias defined as the mean difference between paired BG concentrations obtained with the reference laboratory analyzer and the PBGM. The 95 % CI for the bias of each PBGM was calculated, and the lack of inclusion of 0 in the 95% CI indicated constant bias. The Bland-Altman plots were visually inspected for the presence of proportional bias. Bias and 95% CIs were also calculated for blood samples grouped as hypoglycemic (BG concentration, < 88 mg/dL), normoglycemic (BG concentration, 88 to 135 mg/dL), and hyperglycemic (BG concentration, > 135 mg/dL) to minimize distortion caused by proportional bias.36 Passing-Bablok regression was performed as described elsewhere.37,38 Constant bias was present if the 95% CI for the y-intercept did not include 0. Proportional bias was present if the 95% CI for the slope did not include 1.37 To evaluate differences in agreement between species, a combined absolute bias was calculated for each species. For this, the differences between the reference laboratory analyzer results and the PBGM results were converted to absolute values (ie, no positive or negative signs), summed, and divided by the number of measurements. A species-specific combined absolute bias was considered significant if its 95% CI did not overlap 0. To evaluate changes in agreement associated with differences in Hct, blood samples were categorized into 3 groups according to Hct (low Hct [< 38.0%], high Hct [> 45.0%], and Hct within reference limits [38.0% to 45.0%]). Bias was then calculated with absolute values (ie, no positive or negative signs) and reported with the 95% CI for each Hct category.36 Mean differences, with 95% CIs, were also calculated between results for samples with Hct within reference limits, compared with results for samples with low or high Hct.
As a measure of precision, repeatability of PBGM results was measured by calculating the CV for duplicate measurements.39 Because we found no data in the current literature describing repeatability of the reference method (reference laboratory analyzer [enzymatic hexokinase oxidase reaction]) in Panthera spp, the CV for the reference laboratory analyzer was also calculated. To accomplish this, the sera of 1 tiger and 1 lion were aliquoted in 5 vials each and sent to the laboratory as separate samples for BG concentration analysis. In this way, the laboratory technician was not aware that multiple aliquots of the same sera were being measured.
Total analytic error reflects the sum of random error (imprecision) and systematic error (bias). At the time of the study, no clear consensus existed as to the best method to calculate total analytic error for instrument validation.30,36 Therefore, total error was calculated as described by the ASVCP guidelines40 with the following formula:
Similarly, bias was calculated as follows:
Additionally, on the basis of the ASVCP guidelines, allowable total error for measuring BG concentration was considered to be 10% for hypoglycemic samples, 20% for normoglycemic samples, and 20% for hyperglycemic samples.
For evaluation of diagnostic accuracy of the PBGMs, sensitivity, specificity, and positive and negative predictive values with 95% CIs were calculated. The prespecified BG concentration limits for hypo- and hyperglycemia were < 88 mg/dL and > 135 mg/dL, respectively, as reported for lions < 4 years of age.g These reference limits were used instead of others that had been derived from clinical biochemistry surveys in tigers41,42 or lions42,43 that included only small numbers of individuals or did not report glucose concentrations.
All data obtained, including results from repeated analyses after failure of test strips, were included in the statistical analyses. Data were analyzed with commercial software.h,i For agreement and diagnostic accuracy analyses, the results obtained with the identical PBGMs were averaged. For all other analyses, individual results were used. Results of 2-tailed tests with values of P < 0.05 were considered significant.
Results
Samples
Overall, 53 venous blood samples from 27 tigers and 26 lions were used in the study. Seventeen of the 27 (63%) tigers were males, and 10 (37%) were females. Similarly, 17 of the 26 (65%) lions were males, and 9 (35%) were females. All animals were sexually intact. On the basis of results from the reference laboratory analyzer (reference method), overall mean ± SD BG concentration was 104.2 ± 28.9 mg/dL (median, 100 mg/dL; range, 59 to 218 mg/dL; Table 1). Mean ± SD BG concentration was 109.8 ± 31.8 mg/dL (range, 75 to 218 mg/dL) for the tigers and 98.3 ± 24.9 mg/dL (range, 59 to 161 mg/dL) for the lions. None of the samples had BG concentration outside the analytic range of the PBGMs. Hematocrit was measured in all samples, and overall mean ± SD Hct was 41.3 ± 3.9% (range, 29.8% to 49.6%). Mean ± SD Hct was 41.4 ± 1.8% (range, 39% to 44%) for the tigers and 42.6 ± 4.1% (range, 31% to 47%) for the lions. For tigers and lions combined, mean ± SD BG concentration was 107.5 ± 30.9 mg/dL (range, 70 to 218 mg/dL) for males and 98.2 ± 24.5 mg/dL (range, 59 to 161 mg/dL) for females, and mean ± SD Hct was 41.3 ± 3.3% for males (range, 31% to 45%) and 43.2 ± 3.5% (range, 39% to 47%) for females.
Summary data for BG concentration (mg/dL) measurements obtained with a reference laboratory analyzer, an hPBGM, a vPBGM on canine setting, and a vPBGM on feline setting for 53 venous blood samples from tigers (Panthera tigris; n = 27) and lions (Panthera leo; 26).
| Statistic | Reference laboratory analyzer | hPBGM | vPBGM canine setting | vPBGM feline setting |
|---|---|---|---|---|
| Median (range) | 100.0 (59.0–218.0) | 101.0 (108.5–236.0) | 100.0 (70.0–275.5) | 89.0 (63.0–247.5) |
| Mean ± SD | 104.2 ± 28.9 | 108.5 ± 30.8 | 113.5 ± 39.5 | 101.9 ± 35.5 |
Method agreement
Agreement between the PBGMs and the reference laboratory analyzer varied among instruments (Table 2) but did not vary markedly between species. Combined absolute bias for the 3 PBGMs was 14.2 ± 17.4 mg/dL and 12.8 ± 12.2 mg/dL for the tigers and lions, respectively, and the combined bias for both species was 2.0 mg/dL (95% CI, −2.7 to 6.7 mg/dL). Bland-Altman plots indicated that the vPBGM on canine setting (bias, −9.3 mg/dL; 95% CI, −17.1 to −1.5 mg/dL) had a constant bias, but no constant bias was detected with the hPBGM or the vPBGM on feline setting (Figure 1). In addition, no constant or proportional biases were detected with Passing-Bablok regression analysis (Figure 2).
Bland-Altman agreement plots for BG concentration measurement results obtained with a reference laboratory analyzer and an hPBGM (A), vPBGM on canine setting (B), or vPBGM on feline setting (C) for 53 venous blood samples from tigers (Panthera tigris; n = 27) and lions (Panthera leo; 26). The solid line indicates the mean difference (bias), dashed lines represent the 95% LoA (ie, bias ± 1.96 SD), and dashed and dotted lines represent the 95% CI of the mean difference. Triangles represent individual measurements (determined with the reference laboratory analyzer) with Hct between 38% and 45%, and circles represent individual measurements with Hct < 38% or > 45%. The regression line and 95% CI are depicted to assist detection of proportional bias. RLA = Reference laboratory analyzer.
Citation: Journal of the American Veterinary Medical Association 254, 3; 10.2460/javma.254.3.399
Passing-Bablok regression plots comparing BG concentration measurements obtained with a reference laboratory analyzer and an hPBGM (A), vPBGM on canine setting (B), or vPBGM on feline setting (C) for the blood samples in Figure 1. The solid line represents the linear regression line, and the dotted lines indicate the 95% CI for the regression line. See Figure 1 for remainder of key.
Citation: Journal of the American Veterinary Medical Association 254, 3; 10.2460/javma.254.3.399
Results of Bland-Altman and Passing-Bablok regression analyses of agreement of results obtained with an hPBGM, vPBGM on canine setting, and vPBGM on feline setting, compared with results obtained with a reference laboratory analyzer when measuring BG concentration of blood samples in Table 1.
| Bland-Altman analysis | Passing-Bablok regression | ||||||
|---|---|---|---|---|---|---|---|
| Analyzer | Mean bias (mg/dL)* | 95% CI for mean bias (mg/dL) | 95% LoA (mg/dL) | Regression equation† | 95% CI for y-intercept | 95% CI for slope | CV (%) |
| hPBGM | −4.3 | −10.2 to 1.5 | −46.3 to 37.6 | y = 11.87 + 0.87x | −8.92 to 28.27 | 0.72 to 1.10 | 3.73 |
| vPBGM, canine setting | −9.3 | −17.1 to −1.5 | −64.6 to 46.0 | y = −3.70 + 1.06x | −32.30 to 15.15 | 0.88 to 1.35 | 9.29 |
| vPBGM, feline setting | 2.3 | −4.8 to 9.4 | −47.9 to 52.6 | y = −2.10 + 0.94x | −25.32 to 15.08 | 0.77 to 1.17 | 9.44 |
Calculated as the mean of the differences between results obtained with the reference laboratory analyzer and results obtained with the PBGM for paired samples.
For all PBGMs, no significant (Cusum test, P > 0.05) deviation from linearity was detected.
Effect of Hct
The PBGMs had greater agreement when measuring BG concentration in samples from tigers and lions that had an Hct within reference limits (Hct, 38% to 45%), compared with samples with low or high Hct (Table 3). In addition, mean bias for BG concentration measurements obtained on samples with low or high Hct was 16 mg/dL (95% CI, 4 to 29 mg/dL) greater than mean bias for results obtained on blood samples with an Hct within reference limits when analyzed with the hPBGM, 22 mg/dL (95% CI, 3 to 41 mg/dL) greater when analyzed with the vPBGM on canine setting, and 16 mg/dL (95% CI, 2 to 31 mg/dL) greater when analyzed with the vPBGM on feline setting.
Agreement with reference analyzer values (mean combined absolute bias ± 1.96 SD), precision (CV), and total error for an hPBGM, a vPBGM on canine setting, and a vPBGM on feline setting when measuring BG concentration of blood samples in Table 1 stratified by BG concentration and Hct (below, within, or above reference limits).
| hPBGM | vPBGM, canine setting | vPBGM, feline setting | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Blood sample | No. of animals | Mean bias ± 1.96 SD* (mg/dL) | CV (%) | Total error (%) | Mean bias ± 1.96 SD* (mg/dL) | CV (%) | Total error† (%) | Mean bias ± 1.96 SD* (mg/dL) | CV (%) | Total error† (%) |
| BG concentration (mg/dL) | ||||||||||
| < 88 | 14 | 16.5 ± 19.0 | 2.6 | 22.3‡ | 22.2 ± 26.3 | 8.5 | 38.1‡ | 15.5 ± 22.4 | 8.7 | 29.5‡ |
| 88–135 | 31 | 10.9 ± 12.1 | 3.1 | 7.5 | 12.3 ± 18.6 | 8.4 | 19.3 | 16.6 ± 13.7 | 8.6 | 25.8‡ |
| > 135 | 8 | 25.2 ± 21.3 | 1.8 | 6.3 | 25.0 ± 37.8 | 9.1 | 27.2‡ | 25.8 ± 27.1 | 9.5 | 20.3‡ |
| Hct (%) | ||||||||||
| < 38 | 8 | 24.6 ± 19.3 | 2.9 | — | 30.4 ± 30.1 | 9.0 | — | 28.2 ± 21.8 | 9.3 | — |
| 38–45 | 39 | 10.2 ± 11.7 | 2.6 | — | 11.1 ± 18.5 | 8.4 | — | 13.4 ± 14.2 | 8.6 | — |
| > 45 | 6 | 29.1 ± 24.8 | 3.4 | — | 36.2 ± 36.3 | 9.0 | — | 31.6 ± 29.0 | 9.2 | — |
Calculated by converting the difference between reference analyzer and PBGM values to absolute values (ie, no positive or negative signs).
Total error (%) is reported for various ranges of BG concentrations as indicated by current guidelines.40
None of the PBGMs complied with the maximum allowable total error indicated by the ASVCP guidelines (ie, ≤ 10% for hypoglycemic samples, ≤ 20% for normoglycemic samples, and ≤ 20% for hyperglycemic samples).
— = Total error calculation did not apply to Hct measurements.
Effect of BG concentration
There was no consistent effect of BG concentration on agreement of PBGMs with the reference laboratory analyzer (Table 3). In addition, mean bias for BG concentration measurements obtained on hypo- and hyperglycemic blood samples was 8 mg/dL (95% CI, −0.9 to 18 mg/dL) greater than mean bias for results obtained on normoglycemic samples when analyzed with the hPBGM, 11 mg/dL (95% CI, −4 to 26 mg/dL) greater when analyzed with the vPBGM on canine setting, and 3 mg/dL (95% CI, −9 to 14 mg/dL) greater when analyzed with the vPBGM on feline setting.
Precision
The hPBGM had better overall repeatability (CV, 3.73%) than did the vPBGM on the canine (9.29%) or feline (9.44%) setting (Table 2). The CV of the reference laboratory analyzer was 1.21% for tigers and 1.25% for lions.
Total error
Total error was 11.8% for the hPBGM, 27.7% for the vPBGM on canine setting, and 20.9% for the vPBGM on feline setting. Total error percentages stratified by the glycemic level were calculated (Table 3), and none of the PBGMs, including the vPBGM on feline setting, complied with the maximum allowable total error suggested by the ASVCP guidelines (ie, < 10% total error for hypoglycemic samples, < 20% total error for normoglycemic samples, and < 20% total error for hyperglycemic samples).
Sensitivity and specificity for detecting hypoglycemia and hyperglycemia
Sensitivity and specificity of the PBGMs in detecting hypo- and hyperglycemia in tigers and lions were calculated (Table 4). For detecting hypoglycemia (BG concentration < 88 mg/dL) in tigers and lions, the vPBGM on feline setting had the highest sensitivity (78.6%; 95% CI, 49.2% to 95.3%), whereas the hPBGM and the vPBGM on canine setting had the highest specificity (87.2%; 95% CI, 72.6% to 95.7%). For detecting hyperglycemia (BG concentration > 135 mg/dL) in tigers and lions, the vPBGM on canine setting had the highest sensitivity (87.5%; CI, 47.3% to 99.7%), whereas the hPBGM and the vPBGM on feline setting had the highest specificity (93.3%; CI, 81.7% to 98.6%). However, the 95% CIs for all of these statistics were wide, precluding definitive conclusions.
Sensitivity, specificity, and positive (PPV) and negative (NPV) predictive values of an hPBGM, a vPBGM on canine setting, and a vPBGM on feline setting for diagnosis of hypoglycemia (BG concentration < 88 mg/dL) and hyperglycemia (BG concentration > 135 mg/dL) in 53 blood samples from tigers (n = 27) and lions (26).
| hPBGM | vPBGM, canine setting | vPBGM, feline setting | |||||
|---|---|---|---|---|---|---|---|
| Glycemic status to be detected | Descriptive statistic | Value | 95% CI | Value | 95% CI | Value | 95% CI |
| Hypoglycemia | |||||||
| Sensitivity (%) | 57.1 | 28.9 to 82.3 | 50.0 | 23.0 to 77.0 | 78.6 | 49.2 to 95.3 | |
| Specificity (%) | 87.2 | 72.6 to 95.7 | 87.2 | 72.6 to 95.7 | 71.8 | 55.1 to 85.0 | |
| PPV (%) | 61.5 | 31.6 to 86.1 | 58.3 | 27.7 to 84.8 | 50.0 | 28.2 to 71.8 | |
| NPV (%) | 85.0 | 70.2 to 94.3 | 82.9 | 67.9 to 92.8 | 90.3 | 74.2 to 98.0 | |
| Hyperglycemia | |||||||
| Sensitivity (%) | 62.5 | 24.5 to 91.5 | 87.5 | 47.3 to 99.7 | 75.0 | 34.9 to 96.8 | |
| Specificity (%) | 93.3 | 81.7 to 98.6 | 91.1 | 78.8 to 97.5 | 93.3 | 81.7 to 98.6 | |
| PPV (%) | 62.5 | 24.5 to 91.5 | 63.6 | 30.8 to 89.1 | 66.7 | 29.9 to 92.5 | |
| NPV (%) | 93.3 | 81.7 to 98.6 | 97.6 | 87.4 to 99.9 | 95.4 | 84.5 to 99.4 | |
Blood glucose concentrations obtained with a reference laboratory analyzer were used as the criterion standard.
Discussion
Verification of the performance of PBGMs before clinical implementation has been increasingly recognized as a priority in veterinary medicine,44 especially given that research45 has established that in certain species, the use of particular PBGMs may lead to dramatic clinical errors. For example, hPBGMs have a low specificity (50%) for detecting hypoglycemia in ferrets.45 This means that half of ferrets identified as hypoglycemic with hPBGMs are actually normoglycemic. If we consider that ferrets are prone to hypoglycemia because of insulinomas46 and that hPBGMs were widely used in clinical veterinary practice47 before this low specificity was reported, then this analytic error could have led to innumerable misdiagnoses of hypoglycemia in ferrets during those years. Results of limited precision, agreement, and clinical reliability indicated in the present study suggested that PBGMs currently available for use in veterinary and human medicine are inadequate for measuring BG concentration in tigers and lions.
Performance of the vPBGM used in the present study has been evaluated in ferrets,26 horses,23 alpacas,25 rabbits,27 cats,48,49 and dogs.50 Findings of 1 study48 including 39 cats indicate that results obtained with the vPBGM do not differ significantly from results obtained with a laboratory analyzer; however, no data are provided regarding bias. Another study49 including 85 cats with various diseases shows that the vPBGM overestimates BG concentrations by approximately 3 mg/dL; however, neither the 95% LoA nor the 95% CI of the bias is reported, making comparison with the present study difficult. Nonetheless, 92.9% of the measurements in that study49 are within the International Organization for Standardization acceptance criteria (ie, within 15 mg/dL of the criterion standard when BG concentration is < 75 mg/dL and within 20% of the criterion standard when BG concentration is > 75 mg/dL). Investigators of both of those studies48,49 concluded that the vPBGM is acceptable for clinical use in cats, which is in contrast to results of the present study involving tigers and lions. Although the vPBGM on feline setting, with sensitivity and specificity > 70% for both hypo- and hyperglycemia in the present study, had better clinical reliability, compared with the hPBGM and the vPBGM on canine setting, for measuring BG concentration in tigers and lions, the vPBGM on feline setting also had low precision (CV, 9.44%) and wide 95% LoA, with results that could differ from reference laboratory analyzer results by approximately ± 50 mg/dL. In addition, the poor agreement and precision of each PBGM resulted in a total error for each that was > 20%, higher than current guidelines (ie, ≤ 10% total error for hypoglycemic samples, ≤ 20% total error for normoglycemic samples, and ≤ 20% total error for hyperglycemic samples).
A study50 of dogs shows that 43% of the results obtained from the vPBGM are higher than those of a reference laboratory analyzer, but are not significantly lower or higher than the reference laboratory results. Those findings should be considered cautiously because the bias and 95% CI for the PBGM were not reported, making it difficult to evaluate the significance of the constant overestimation of BG concentration observed.35,51 More recently, the vPBGM has been shown to have proportional and constant bias in dogs.49 In ferrets, the vPBGM on canine setting had a small bias (1.9 mg/dL) but wide 95% LoA (−29 to 34 mg/dL; values extrapolated from the Bland-Altman plot)26 that could affect clinical decision-making, as also stated by the authors.26 However, when the vPBGM on feline setting was used in ferrets, underestimation of BG concentration was observed.26 In rabbits, a substantial overestimation of BG concentration occurred with the vPBGM on canine setting (bias, 48 mg/dL; 95% LoA, −20 to 116 mg/dL) and feline setting (bias, 25 mg/dL; 95% LoA, −27 to 78 mg/dL) with the same methods used.27 Results of that study27 show that this overestimation would have a clinical impact because use of the vPBGM on canine or feline setting would result in misdiagnosis of hyperglycemia in 9% and 5.6% of normoglycemic rabbits, respectively, and misdiagnosis of normoglycemia in 1.1% of hyperglycemic rabbits. As in rabbits, the vPBGM typically overestimates alpaca BG concentration (bias, 5.2 mg/dL) and has a wide 95% LoA (−46.6 to 57.1 mg/dL).25 In horses, however, constant over- or underestimation of BG concentration when measured with the vPBGM has not been reported, and the vPBGM is considered clinically acceptable, with nearly 97% of the results contributing to appropriate clinical decisions in horses.23 Given these interspecies differences, it is key to assess the performance of veterinary PBGMs in the target species. Use of a common statistical analysis technique and study design across species would facilitate better overall interpretation and meta-analysis of results.35,40,52
Results of the present study indicated that the vPBGM on canine setting had the worst precision, agreement, and clinical reliability for detecting hypoglycemia in tigers and lions, but had acceptable sensitivity and specificity for detecting hyperglycemia in both species. The hPBGM (CV, 3.73%) had approximately 6% better repeatability than the vPBGM on canine (9.29%) or feline (9.44%) settings in the present study. Interestingly, the hPBGM also has better repeatability (CV, 3%) than does the vPBGM (CV, 4%) when measuring BG concentration in rabbits.27 Results of the present study, however, indicated that clinical use of the hPBGM in tigers and lions would not be recommended because 95% of results from the hPBGM could differ from the true BG concentration by approximately ± 40 mg/dL and because sensitivity for detecting hypo- and hyperglycemia was 57.1% and 62.5%, respectively. Curiously, results for the hPBGM in the present study did not indicate a constant underestimation of BG concentration in tigers and lions, as observed in horses,53 parrots,18 alpacas,20,25,45 ferrets,26 and rabbits.27 This lack of a constant underestimation of BG concentration by the hPBGM in tigers and lions in the present study was similar to the proportional bias (−2.7 mg/dL; 95% LoA, −31.3 to 26.1 mg/dL) observed when the instrument is used in humans.10,54
The poor performances of PBGMs evaluated in the present study were likely multifactorial, and the present study was not designed to specifically determine why the PBGMs did not have good agreement with the reference laboratory analyzer. Nonetheless, we suspected that the filters used in the test strips to separate erythrocytes from plasma could have contributed to this disagreement53 because the mean corpuscular volume for erythrocytes in tigers and lions is approximately 50 to 65 fL55 and 50 to 52 fL,43,56 respectively, whereas that for humans and dogs is approximately 80 to 100 fL57 and 60 to 77 fL,58 respectively. Therefore, the filters in the test strips may have allowed passage of the smaller erythrocytes of tigers and lions, potentially interfering with PBGM results.
Hemorheologic and hemodynamic properties of blood could also have been factors in the poor performance of these PBGMs. Considering that BG concentration is measured in the aqueous portion of a blood sample and that serum and plasma have higher water content and BG concentration than does whole blood,59 BG concentration results obtained when analyzing plasma or serum with a reference laboratory analyzer are expected to be higher than results obtained when analyzing whole blood with a PBGM. For instance, plasma and the packed cell component of blood in humans are 93% and 71% water by volume, respectively,60 and when mean Hct of 43% is considered, the constant factor to convert molality of glucose in whole blood to the equivalent molarity of glucose in plasma is 1.11.61 In addition, it is expected that a fixed volume of plasma has approximately 11% to 12% higher water content and glucose concentration, compared with whole blood that has an Hct of approximately 45%.31 Some PBGMs have calibrations to correct this incongruity, assuming the sample has an Hct within a given interval.59
Species differences in distribution of glucose in plasma and erythrocytes may have played a role in the discrepancies between results for the reference laboratory analyzer and the PBGMs in the present study.23 For instance, in humans, approximately 60% of BG is located in erythrocytes and 40% is in plasma,62 whereas in cats, only about 7% of BG is located in erythrocytes and 93% is in plasma.63
Several factors could have adversely affected BG concentration results obtained with the PBGMs.61 For instance, the accuracy of the PBGMs could have been affected by sample Hct64,65 because blood samples with an extremely high Hct may have had an altered viscosity, which could have altered the diffusion kinetics or have had insufficient plasma volume for accurate BG concentration measurement with a PBGM.66 In the present study, the PBGMs had poorer agreement when used on samples with Hct values outside of reference limits. In dogs and rabbits, Hct influences the accuracy of PBGMs but in an instrument-dependent way, such that for samples with a high Hct, the hPBGM increases the underestimation of BG concentration, whereas for samples with a low Hct, the vPBGM increases the overestimation of BG concentration.27,28 Therefore, findings from those studies27,28 indicate that the hPBGM is likely to be less accurate in polycythemic animals and more accurate in anemic ones, whereas the vPBGM is likely to be more accurate in polycythemic animals and less accurate in anemic ones. In cats, however, the same vPBGM has lower accuracy when analyzing samples with a high Hct (> 48%), compared with samples that have an Hct within reference limits.49
In addition to the Hct, the anatomic site used for blood collection may have affected BG concentration results of the PBGMs because these instruments are designed to measure BG concentration in capillary blood, whereas venous blood was used in the present study. Although we could not be certain that repeating the study with capillary blood samples instead would have yielded identical results, no clinically important differences in results were reported67 for cats when capillary samples were obtained from the marginal ear vein, compared with results of venous samples. Considering that all blood samples evaluated in the present study were of venous origin, all instruments used in the present study should have been affected similarly.
Apart from the Hct and anatomic site sampled, analytes in the blood other than BG could have affected agreement between the PBGMs and the reference laboratory analyzer. For instance, when assessed in humans, PBGMs performed well as long as the Hct was within reference limits and no maltose or ascorbate was present; however, total analytic error approached 30% when blood with an Hct of 20% and an ascorbate concentration of 0.29 mmol/L was evaluated.61 Future studies should consider these sources of analytic error when PBGMs are evaluated for use in animals.
A limitation of the present study was that it did not assess all factors (eg, differences between BG concentration in plasma and whole blood44) that could be evaluated. However, the present study was designed to mimic clinical use of PBGMs in field settings, and in such field settings, we expected that analysis of whole blood would be more convenient, compared with analysis of plasma or serum that would first require centrifugation. Another limitation was that interference by other analytes (eg, maltose and ascorbate61) was not evaluated. Different concentrations of such analytes in tigers and lions, compared with humans and cats, could explain the poor agreement observed in the present study. A further limitation was the use of a BG concentration reference range that had not been generated with receiver operating characteristic curves in diagnostic accuracy studies that included healthy and diseased animals. Because such research was lacking in tigers and lions at the time of the present study, we used results from a surveyg of glucose concentrations in apparently healthy lions. Finally, the interpretation of sensitivity and specificity for the PBGMs was limited owing to their wide CIs calculated in the present study.
Collectively, results of the present study indicated that performance of the evaluated PBGMs was inadequate for measuring BG concentration in tigers and lions. This conclusion was supported by the limited precision, agreement, and clinical reliability of the PBGMs evaluated. Although variations in Hct played a role in disagreement between results from the PBGMs and results from the reference laboratory analyzer, further work is required to fully understand the causes of the disagreement when measuring BG concentration in tigers and lions.
Acknowledgments
The authors had no financial interests with companies that manufactured products used in the present study or with companies that manufactured competing products.
ABBREVIATIONS
| ASVCP | American Society for Veterinary Clinical Pathology |
| BG | Blood glucose |
| CI | Confidence interval |
| CV | Coefficient of variation |
| hPBGM | Portable blood glucose meter designed for use in humans |
| LoA | Limits of agreement |
| PBGM | Portable blood glucose meter |
| POC | Point-of-care |
| vPBGM | Veterinary portable blood glucose meter designed for use in dogs and cats |
Footnotes
Accu-Chek Aviva, Roche Diabetese Care Inc, Indianapolis, Ind.
AlphaTRAK 2 BC monitoring system, Abbott Laboratories, Abbott Park, Ill.
Saturno 300, Crony Instruments Srl, Roma, Italy.
AccuChek Aviva test strips, Roche Diagnostics GmbH, Mannheim, Germany.
AlphaTRAK 2 BC test strips, Abbott Laboratories, Abbott Park, Ill.
Algorithm, Abbott Laboratories, Abbott Park, Ill.
Erasmus HL. Determination of some blood parameters in the African lion (Panthera leo). MS dissertation, University of the Free State, Bloemfontein, South Africa, 2008.
MedCalc, version 12.2.1, MedCalc Software bvba, Mariakerke, Belgium.
SPSS, version 22.0, IBM Corp, Chicago, Ill.
References
1. Wintergerst KA, Buckingham B, Gandrud L, et al. Association of hypoglycemia, hyperglycemia, and glucose variability with morbidity and death in the pediatric intensive care unit. Pediatrics 2006;118:173–179.
2. Torre DM, deLaforcade AM, Chan DL. Incidence and clinical relevance of hyperglycemia in critically ill dogs. J Vet Intern Med 2007;21:971–975.
3. Hassel DM, Hill AE, Rorabeck RA. Association between hyperglycemia and survival in 228 horses with acute gastrointestinal disease. J Vet Intern Med 2009;23:1261–1265.
4. Swain JM, Pirie RS, Hudson NP, et al. Insulin-like growth factors and recurrent hypoglycemia associated with renal cell carcinoma in a horse. J Vet Intern Med 2005;19:613–616.
5. Elfenbein J, Credille B, Camus M, et al. Hypoglycemia and hyperlactatemia associated with lymphoma in an Angus cow. J Vet Intern Med 2008;22:1441–1443.
6. Lumeij JT, van der Hage MH, Dorrestein GM, et al. Hypoglycaemia due to a functional pancreatic islet cell tumour (insulinoma) in a ferret (Mustela putorius furo). Vet Rec 1987;120:129–130.
7. Cello RM, Kennedy PC. Hyperinsulinism in dogs due to pancreatic islet cell carcinoma. Cornell Vet 1957;47:538–557.
8. Reilly S, Seddighi MR, Steeil JC, et al. Selected clinical, biochemical, and electrolyte alterations in anesthetized captive tigers (Panthera tigris) and lions (Panthera leo). J Zoo Wildl Med 2014;45:328–334.
9. Wess G, Reusch C. Capillary blood sampling from the ear of dogs and cats and use of portable meters to measure glucose concentration. J Small Anim Pract 2000;41:60–66.
10. Tack C, Pohlmeier H, Behnke T, et al. Accuracy evaluation of five blood glucose monitoring systems obtained from the pharmacy: a European multicenter study with 453 subjects. Diabetes Technol Ther 2012;14:330–337.
11. ASVCP Quality Assurance and Laboratory Standards (QALS) Committee, Flatland B, Freeman KP, et al. ASVCP guidelines: quality assurance for point-of-care testing in veterinary medicine. Available at: www.asvcp.org/pubs/qas/newQas/PDF/ASVCP%20POCT%20QA%20Guideline%20May%202013.FINAL.pdf. Accessed Aug 5, 2013.
12. Joseph RJ, Allyson K, Graves TK, et al. Evaluation of two reagent strips and three reflectance meters for rapid determination of blood glucose concentrations. J Vet Intern Med 1987;1:170–174.
13. Cohn LA, McCaw DL, Tate DJ, et al. Assessment of five portable blood glucose meters, a point-of-care analyzer, and color test strips for measuring blood glucose concentration in dogs. J Am Vet Med Assoc 2000;216:198–202.
14. Burdick S, Mitchell MA, Neil J, et al. Evaluation of two point-of-care meters and a portable chemistry analyzer for measurement of blood glucose concentrations in juvenile white-tailed deer (Odocoileus virginianus). J Am Vet Med Assoc 2012;240:596–599.
15. Olateju T, Begley J, Flanagan D, et al. Effects of simulated altitude on blood glucose meter performance: implications for in-flight blood glucose monitoring. J Diabetes Sci Technol 2012;6:867–874.
16. Clarke SF, Foster JR. A history of blood glucose meters and their role in self-monitoring of diabetes mellitus. Br J Biomed Sci 2012;69:83–93.
17. Lieske CL, Ziccardi MH, Mazet JAK, et al. Evaluation of 4 handheld blood glucose monitors for use in seabird rehabilitation. J Avian Med Surg 2002;16:277–285.
18. Acierno MJ, Mitchell MA, Schuster PJ, et al. Evaluation of the agreement among three handheld blood glucose meters and a laboratory blood analyzer for measurement of blood glucose concentration in Hispaniolan Amazon parrots (Amazona ventralis). Am J Vet Res 2009;70:172–175.
19. Katsoulos PD, Minas A, Karatzia MA, et al. Evaluation of a portable glucose meter for use in cattle and sheep. Vet Clin Pathol 2011;40:245–247.
20. Beemer O, Byers S, Bohn A. Evaluation of four point-of-care glucose meters in alpacas. J Vet Intern Med 2013;27:990–995.
21. Wess G, Reusch C. Evaluation of five portable blood glucose meters for use in dogs. J Am Vet Med Assoc 2000;216:203–209.
22. Wess G, Reusch C. Assessment of five portable blood glucose meters for use in cats. Am J Vet Res 2000;61:1587–1592.
23. Hackett ES, McCue PM. Evaluation of a veterinary glucometer for use in horses. J Vet Intern Med 2010;24:617–621.
24. Wittrock JA, Duffield TF, Leblanc SJ. Short communication: validation of a point-of-care glucometer for use in dairy cows. J Dairy Sci 2013;96:4514–4518.
25. Tennent-Brown BS, Koenig A, Williamson LH, et al. Comparison of three point-of-care blood glucose meters for use in adult and juvenile alpacas. J Am Vet Med Assoc 2011;239:380–386.
26. Petritz OA, Antinoff N, Chen S, et al. Evaluation of portable blood glucose meters for measurement of blood glucose concentration in ferrets (Mustela putorius furo). J Am Vet Med Assoc 2013;242:350–354.
27. Selleri P, Di Girolamo N, Novari G. Performance of two portable meters and a benchtop analyzer for blood glucose concentration measurement in rabbits. J Am Vet Med Assoc 2014;245:87–98.
28. Paul AE, Shiel RE, Juvet F, et al. Effect of hematocrit on accuracy of two point-of-care glucometers for use in dogs. Am J Vet Res 2011;72:1204–1208.
29. Dobromylskyj MJ, Sparkes AH. Assessing portable blood glucose meters for clinical use in cats in the United Kingdom. Vet Rec 2010;167:438–442.
30. Jensen AL, Kjelgaard-Hansen M. Method comparison in the clinical laboratory. Vet Clin Pathol 2006;35:276–286.
31. Burtis CA, Ashwood ER. Tietz textbook of clinical chemistry. 3rd ed. Philadelphia: WB Saunders Co, 1999.
32. Baumstark A, Pleus S, Schmid C, et al. Lot-to-lot variability of test strips and accuracy assessment of systems for self-monitoring of blood glucose according to ISO 15197. J Diabetes Sci Technol 2012;6:1076–1086.
33. AccuChek Aviva user's manual. Mannheim, Germany: Roche Diagnostics GmbH, 2014.
34. AlphaTRAK user guide. North Chicago, Ill: Abbott Laboratories, 2008.
35. Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986;1:307–310.
36. Gerber KL, Freeman KP, ASVCP Quality Assurance and Laboratory Standards (QALS) Committee. ASVCP guidelines: quality assurance for portable blood glucose meter (glucometer) use in veterinary medicine. Available at: www.asvcp.org/about/committees/ASVCP_Guidelines_for_GlucoseMeter.pdf. Accessed Dec 21, 2016.
37. Bablok W, Passing H, Bender R, et al. A general regression procedure for method transformation. Application of linear regression procedures for method comparison studies in clinical chemistry, part III. J Clin Chem Clin Biochem 1988;26:783–790.
38. Passing H. Bablok W. A new biometrical procedure for testing the equality of measurements from two different analytical methods. Application of linear regression procedures for method comparison studies in clinical chemistry, part I. J Clin Chem Clin Biochem 1983;21:709–720.
39. Jones R, Payne B. Clinical investigation and statistics in laboratory medicine. London: CB Venture Publications, 1997.
40. ASVCP Quality Assurance and Laboratory Standards (QALS) Committee, Harr KE, Flatland B, et al. ASVCP guidelines: allowable total error. Available at: www.asvcp.org/about/committees/pdf/ASVCP_Allowable_Total_Error_Recommendations-Biochemistry.pdf. Accessed Aug 5, 2013.
41. Shrivatav AB, Singh KP, Mittal SK, et al. Haematological and biochemical studies in tigers (Panthera tigris tigris). Eur J Wildl Res 2012;58:365–367.
42. Larsson MHMA, Santo PDE, Mirandola RMS. Hematologic parameters of captive lions (Panthera leo) and Siberian tigers (Panthera tigris altaica). Acta Sci Vet 2015;43:1311.
43. Maas M, Keet DF, Nielen M. Hematologic and serum chemistry reference intervals for free-ranging lions (Panthera leo). Res Vet Sci 2013;95:266–268.
44. Brito-Casillas Y, Figueirinhas P, Wiebe JC, et al. ISO-based assessment of accuracy and precision of glucose meters in dogs. J Vet Intern Med 2014;28:1405–1413.
45. Summa NM, Eshar D, Lee-Chow B, et al. Comparison of a human portable glucometer and an automated chemistry analyzer for measurement of blood glucose concentration in pet ferrets (Mustela putorius furo). Can Vet J 2014;55:865–869.
46. Li X, Fox JG, Padrid PA. Neoplastic diseases in ferrets: 574 cases (1968–1997). J Am Vet Med Assoc 1998;212:1402–1406.
47. Lewington J. Ferret husbandry, medicine and surgery. 2nd ed. Oxford, England: Saunders Ltd, 2007.
48. Zini E, Moretti S, Tschuor F, et al. Evaluation of a new portable glucose meter designed for the use in cats. Schweiz Arch Tierheilkd 2009;151:448–451.
49. Kang MH, Kim D-H, Jeong I-S, et al. Evaluation of four portable blood glucose meters in diabetic and non-diabetic dogs and cats. Vet Q 2016;36:2–9.
50. Cohen TA, Nelson RW, Kass PH, et al. Evaluation of six portable blood glucose meters for measuring blood glucose concentration in dogs. J Am Vet Med Assoc 2009;235:276–280.
51. Johnson R. Assessment of bias with emphasis on method comparison. Clin Biochem Rev 2008;29(suppl 1):S37–S42.
52. Zaki R, Bulgiba A, Ismail R, et al. Statistical methods used to test for agreement of medical instruments measuring continuous variables in method comparison studies: a systematic review. PLoS One 2012;7:e37908.
53. Hollis AR, Dallap Schaer BL, Boston RC, et al. Comparison of the Accu-Chek Aviva point-of-care glucometer with blood gas and laboratory methods of analysis of glucose measurement in equine emergency patients. J Vet Intern Med 2008;22:1189–1195.
54. Zueger T, Schuler V, Stettler C, et al. Assessment of three frequently used blood glucose monitoring devices in clinical routine. Swiss Med Wkly 2012;142:w13631.
55. Sajjad S, Farooq U, Malik H, et al. Comparative hematological variables of Bengal tigers (Panthera tigris tigris) kept in Lahore Zoo and Lahore Wildlife Park, Pakistan. Turk J Vet Anim Sci 2012;36:346–351.
56. Teare JA. Physiological data reference values [CD-ROM]. Apple Valley, Minn: International Species Information System, 2002.
57. Giorno R, Clifford JH, Beverly S, et al. Hematology reference values. Analysis by different statistical technics and variations with age and sex. Am J Clin Pathol 1980;74:765–770.
58. Handagama P, Feldman B, Kono C, et al. Mean platelet volume artifacts: the effect of anticoagulants and temperature on canine platelets. Vet Clin Pathol 1986;15:13–17.
59. Tonyushkina K, Nichols JH. Glucose meters: a review of technical challenges to obtaining accurate results. J Diabetes Sci Technol 2009;3:971–980.
60. D'Orazio P, Burnett RW, Fogh-Andersen N, et al. Approved IFCC recommendation on reporting results for blood glucose: International Federation of Clinical Chemistry and Laboratory Medicine Scientific Division, Working Group on Selective Electrodes and Point-of-Care Testing (IFCC-SD-WG-SEPOCT). Clin Chem Lab Med 2006;44:1486–1490.
61. Lyon ME, DuBois JA, Fick GH, et al. Estimates of total analytical error in consumer and hospital glucose meters contributed by hematocrit, maltose, and ascorbate. J Diabetes Sci Technol 2010;4:1479–1494.
62. Andreen-Svedberg A. On the distribution of sugar between plasma and corpuscles in animal and human blood. Skand Arch Physiol 1933;66:113–190.
63. Coldman MF, Good W. The distribution of sodium, potassium and glucose in the blood of some mammals. Comp Biochem Physiol 1967;21:201–206.
64. Tang Z, Lee JH, Louie RF, et al. Effects of different hematocrit levels on glucose measurements with handheld meters for point-of-care testing. Arch Pathol Lab Med 2000;124:1135–1140.
65. Karon BS, Griesmann L, Scott R, et al. Evaluation of the impact of hematocrit and other interference on the accuracy of hospital-based glucose meters. Diabetes Technol Ther 2008;10:111–120.
66. Louie RF, Tang Z, Sutton DV, et al. Point-of-care glucose testing: effects of critical care variables, influence of reference instruments, and a modular glucose meter design. Arch Pathol Lab Med 2000;124:257–266.
67. Thompson MD, Taylor SM, Adams VJ, et al. Comparison of glucose concentrations in blood samples obtained with a marginal ear vein nick technique versus from a peripheral vein in healthy cats and cats with diabetes mellitus. J Am Vet Med Assoc 2002;221:389–392.
