Evaluation of a point-of-care device for creatinine measurement in nonazotemic pigs presented to a referral hospital

Abstract

Objective

To compare creatinine concentrations measured using a point-of-care unit (Nova StatSensor; NSS) to those from a comparison method (CM) chemistry analyzer from pigs presented at a referral institution.

Methods

A clinical prospective study was performed including 61 pigs admitted to Cornell Nemo Farm Animal Hospital for clinical reasons. Heparinized blood samples were obtained by venous aspiration, and whole-blood samples were obtained via ear prick. Creatinine concentrations were subsequently measured in triplicate on the NSS (on heparinized venous samples [NSS_ven] and ear prick samples [NSS_ear]) and compared to the CM. Paired t tests and intraclass correlation coefficients were performed. A Bland-Altman plot was graphed based on a linear mixed-effects model to estimate agreement between tests. Equivalence testing was performed.

Results

NSS_ven (n = 110) and NSS_ear (71) from 61 nonazotemic pigs were analyzed. The mean creatinine ± SD was 1.3 ± 0.52 mg/dL from the CM, 1.1 ± 0.43 mg/dL from the NSS_ven, and 1.2 ± 0.42 mg/dL from the NSS_ear. Bias existed between average NSS_ven (−0.17 mg/dL) and NSS_ear (−0.21 mg/dL) and the CM. The 95% CI of the limit of agreement between the NSS_ven and CM was −0.88 to 0.54 and −1.04 to 0.62 between the NSS_ear and CM. The tests were not equivalent at 0.3 mg/dL.

Conclusions

The NSS_ven, NSS_ear, and CM results are different and not equivalent in nonazotemic pigs.

Clinical Relevance

The use of this device for screening of creatinine in pigs is discouraged based on the results in this population.

Plasma creatinine concentrations are a key indicator of renal function.¹ Both screening and serial monitoring of creatinine are used to guide clinical decisions related to overall management of veterinary patients, including drug selection and fluid therapy. Clinicians may use point-of-care units to assess creatinine concentrations due to the rapid availability of results.

The Nova StatSensor (NSS; Nova Biomedical Corp) is a point-of-care device that uses 20 μL of whole blood to measure creatinine via multiple enzyme reactions (Figure 1). The NSS has been evaluated in human patients with renal injury, and creatinine measurements from whole blood correlated well with comparison method (CM) plasma creatinine testing using modified kinetic Jaffe assay and enzymatic colorimetric creatinine method.^1,2 Due to these promising results, the NSS has been recently evaluated in veterinary species, such as goats and dogs. The NSS adequately identified normal and mildly azotemic patients and had an acceptable accuracy of creatinine measurement in goats.³ However, the device was found unacceptable for clinical use in dogs as it was found not reliable in azotemic patients.⁴ Inconsistent performance of the NSS indicates the importance of species-specific test evaluation.

Figure 1—

The Nova StatSensor (NSS) handheld device with reagent strip in place for blood creatinine concentration measurement.

The growing popularity of pigs as companion pets in the US has led to a rise in porcine elective and emergency care veterinary visits.⁵ Companion pig medicine focuses on individual conditions versus herd-based problems, necessitating medical approaches tailored for the individual patient. Blood sampling is integral to patient evaluation and monitoring. However, reliable venous access is a significant challenge in both healthy and sick pigs due to their unique anatomy and temperament.⁶ As an added challenge, pigs have higher coagulability compared to other species.^7,8 For this reason, the risk of sample inadequacy increases if blood sampling is prolonged or traumatic or with multiple venipuncture attempts. Ear-prick sampling provides a potential alternative to venipuncture for point-of-care creatinine measurements in pigs. As in other species, quick test turnaround and accurate blood value measurements are useful in the assessment and case management of critically ill and stable pigs.⁹ Point-of-care devices, such as the NSS, which require minimal blood volumes, could offer this option.⁹ No commercially available, handheld point-of-care creatinine concentration analyzers have been evaluated for use in pigs. The NSS, which can detect creatinine between 0.3 and 12 mg/dL per manufacturing guidelines, has the potential to be a valuable tool in pig companion medicine. Therefore, the aim of this study was to assess the precision of the NSS when porcine samples are obtained by venous aspiration and ear prick; additionally, an objective was to evaluate the equivalence and agreement of the NSS to a laboratory chemistry analyzer. The NSS results are hypothesized to be precise and equivalent to those of the CM analyzer, as assessed using a linear mixed-effects model, and equivalence testing.

Methods

Animal use and study enrollment

A clinical prospective study was performed from January through July 2025. Any pig admitted to Cornell University Farm Animla Hospital that underwent blood collection (unsedated, sedated, or anesthetized) or had IV catheterization was eligible for inclusion in the study. Cornell University Insitutional Animal Care and Use Commitee (IACUC) approved the procedures and use of live animals for this study (IACUC protocol No. 2024–0178). Owner consent was obtained prior to patient enrollment. The data recorded included patient signalment, reason for hospitalization, and blood collection method (venipuncture, IV catheter aspiration, ear prick).

Sample handling and testing

Quality-control testing of the NSS was performed once every 24 hours when in use according to manufacturer guidelines. Device errors encountered during sample analysis were recorded. Approximately 1 mL of whole blood was collected per sampling episode by venipuncture or venous catheter aspiration and placed into a lithium heparin blood tube (BD Vacutainer; Becton, Dickinson and Co). When blood was collected from a venous catheter, the sample was collected via “3-syringe technique.” The catheter was flushed with heparinized saline, presample blood (> 1 mL) was aspirated to clear the catheter of fluid (remaining crystalloid or heparinized saline), and then an undiluted sample was collected and placed into lithium heparin blood tubes for analysis. The presample blood was discarded or returned to the patient immediately as long as no clotting was visualized, and the catheter was flushed with heparinized saline. Heparinized venous blood was tested on the NSS in triplicate (NSS_ven), and the same sample was submitted to the Clinical Pathology Laboratory at Cornell University for creatinine concentrations measurement via CM (Cobas 501c analyzer; Roche). Ear-prick sampling (NSS_ear) was also performed at the time of venous sample collection when patient cooperation allowed. Blood obtained from NSS_ear was also tested in triplicate by direct contact between the NSS test strip and the blood from the auricular surface following a needle prick. The NSS measures creatinine by converting creatinine to ferricyanide and quantifying the resulting current via amperometry. The CM uses a 2-point enzymatic reaction to convert creatinine to a chromogenic quinone and measures its concentration by colorimetry. Venipuncture, or venous catheter aspiration, and NSS_ear (Figure 2) were performed a maximum of 3 times/patient/hospital visit. Pigs that remained hospitalized or returned for multiple visits were eligible for reenrollment and repeat of the sampling procedure for a maximum of 3 times/visit. A maximum of 6 total sample time points/pig were allowed.

Figure 2—

Example of how ear pricks were performed during the study: NSS, with strip inserted, applied to the auricular surface following a needle prick.

Statistical analysis

Population sampling size was determined by a priori power analysis using ClinCalc sample size calculator (ClinCalc LLC, Chicago, IL, USA; updated June 23, 2024; available at https://clincalc.com/stats/samplesize.aspx). Setting the α at .05 and power at 0.8, a minimum of 32 samples was found necessary to detect a statistically significant creatinine concentration difference of at least 0.3 mg/dL between the NSS and the CM. Descriptive statistics (Excel, version 16.0; Microsoft Corp) were used to summarize patient signalment, reason for hospitalization, and device errors. Data normality was assessed visually with quantile-quantile (Q-Q) plots and histograms. Normally distributed results were reported as mean and SD, and non–normally distributed data were reported as median and range. Paired t tests were performed (Prism, version 9; GraphPad Software Inc) between the mean of the triplicate measurements, from NSS_ven, NSS_ear, and the CM. An intraclass correlation coefficient (ICC) was calculated to evaluate repeatability among triplicate measurements of creatinine on the NSS_ven and NSS_ear, respectively. The ICC was calculated using RStudio (version 4.4.2; Posit PBC) using a 2-way mixed-effects model for consistency with single measurements (ICC [3,1]). The ICC was interpreted as excellent (0.91 to 1), good (0.76 to 0.9), moderate (0.5 to 0.75), or poor (< 0.5). A linear mixed-effects model was run, and a Bland-Altman adapted for repeated measures was plotted using the bias from the linear mixed-effects model as the upper and lower bounds.

Nova StatSensor averaged values between the 3 replicates were utilized in the model. The model was run with a fixed effect representing mean bias and a random intercept for subject; the response variable was the difference between the NSS and CM. The random effect was pig ID to account for repeated measurements. Variance from the model was used to estimate 95% limits of agreement, accounting for within-subject replicates.

An equivalence test between the NSS and CM was run with bounds of equivalence set at 0.3 mg/dL. The 0.3-mg/dL bound was elected due to its clinical importance as this absolute increase correlates to acute kidney injury.¹⁰ Significance was set at a P value of < .05.

Results

Normality testing showed large right skewing of the data on the histogram and deviation from the Q-Q line on the Q-Q plot due to a single patient. Removal of this individual demonstrated normal distribution on both the histogram and Q-Q plot. Because this individual represented an extreme outlier, they were removed from further analysis.

Pigs (n = 61) of various breeds, age, presentation status, and reason for hospitalization were included in the study and are reported in Supplementary Tables S1 and S2.

No machine or strip errors were encountered when most samples (143 of 181 [79%]) were processed. However, at least 1 read error was encountered in 38 of 181 samples (21%). Errors most frequently occurred with NSS_ear (23 of 38 [60%]). The number of errors ranged from 1 to 12, with a median of 1/sampling time point.

Venipuncture or venous catheter samples (n = 110) and NSS_ear (71) were collected from enrolled patients. Sites of blood collection and catheterization included coccygeal, auricular, cephalic, jugular, lateral saphenous, and superficial epigastric veins. Fifty-three pigs were enrolled during 1 visit only, 7 pigs were enrolled during 2 separate visits, and 1 pig was enrolled at each of its 4 visits. Two pigs were enrolled a total of 4 times; all of the other pigs were enrolled 3 or fewer times.

The mean NSS_ven and NSS_ear creatinine concentrations ± SD were 1.1 ± 0.43 mg/dL and 1.2 ± 0.42 mg/dL, respectively. The mean creatinine concentration ± SD from the CM measurements was 1.3 ± 0.52 mg/dL. The single outlier patient identified and removed presented with these values: CM, 10.6 mg/dL; NSS_ven mean, 6.41 ± 0.4 mg/dL; and NSS_ear mean, 5.05 ± 1.59 mg/dL. No other values above 2.4 mg/dL were identified in the dataset, limiting the evaluation of the NNS performance to nonazotemic samples. Table 1 reports the mean creatinine measurements for the CM and NSS based on patient presentation.

Table 1—

Mean and range of creatinine concentrations measured by the comparison method (laboratory chemical analyzer) and the Nova StatSensor method, grouped by patient presentation.

Presentation		Comparison method	Nova StatSensor venous aspiration	Nova StatSensor ear prick
Emergency	Mean (mg/dL)	1.31	1.07	1.31
20/71 (28%)	Range (mg/dL)	0.30–2.30	0.43–2.07	0.53–2.11
	Animals	20	19	11
Elective	Mean (mg/dL)	1.28	1.17	1.21
51/71 (72%)	Range (mg/dL)	0.60–2.40	0.41–2.38	0.60–2.02
	Animals	51	51	36

A significant difference between the NSS mean creatinine and CM creatinine was detected for both NSS_ven and NSS_ear, respectively (P < .0001 for all). No significant difference was detected between mean creatinine from NSS_ven and mean creatinine from NSS_ear (P = .63). When evaluating repeatability of the NSS values, the ICC demonstrated good correlation among triplicate values from samples acquired by NSS_ven (ICC, 0.80; 95% CI, 0.74 to 0.85; P < .0001). Moderate correlation was detected among triplicate NSS_ear samples (ICC, 0.75; 95% CI, 0.66 to 0.83; P < .0001).

Using the linear mixed-effects model comparing NSS_ven and the CM, the bias was −0.17 mg/dL (95% CI, −0.88 to 0.54 mg/dL), meaning the NSS_ven, on average, read 0.17 mg/dL lower than the CM. The between-patient SD was 0.23 mg/dL, and the within-patient SD was 0.28 mg/dL. The bias and 95% CI were used to graph the mean and upper and lower limits of agreement for the Bland-Altman plot (Figure 3). Using the linear mixed-effects model comparing NSS_ear and the CM values, the bias was −0.21 mg/dL (95% CI, −1.04 to 0.62 mg/dL), meaning the NSS_ear, on average, read 0.21 mg/dL lower than the CM. The between-patient SD was 0.26 mg/dL, and the within-patient SD was 0.33 mg/dL. The bias and 95% CI were used to graph the mean and upper and lower limits of agreement for the Bland-Altman plot. In both Bland-Altman plots, a wide distribution of values, with several falling outside the established CI, was detected.

Figure 3—

A and B—Bland-Altman plots from the linear mixed-effects model obtained from values of the NSS measurements by venipuncture/venous catheter aspiration and the comparison method (CM; A) and between the NSS values obtained via ear prick and the CM (B). The solid center line indicates the mean difference, and the dotted lines represent the limits of agreement (mean ± 1.96 SD) obtained from the linear mixed model. The x-axis represents the mean (mg/dL), and the y-axis represents the difference (mg/dL) between the NSS means and the CM values.

The creatinine concentrations measured by the NSS_ven as well as the NSS_ear compared to the CM were not equivalent at 0.3 mg/dL (P = .86 and P = .76, respectively).

Discussion

This study evaluated the NSS agreement and equivalence to the CM analyzer when measuring blood creatinine in nonazotemic pigs presenting for elective and emergent evaluation at Cornell Nemo Farm Animal Hospital. Intratest repeatability indicated that the NSS produced good correlation between 3 measurements of the same sample when blood was obtained by venipuncture/venous catheter aspiration and moderate correlation between measurements when blood was obtained by ear prick. As detected by the t test, a significant difference exists between the results generated by the NSS, regardless of the sample type, and the CM analyzer.

The detected difference between the NSS and CM could be relevant in clinical settings; therefore, to better assess the clinical usefulness of the handheld device with either sample type, a linear mixed-effects model was estimated. Based on the linear model, negative bias indicates that the NSS underestimates the value compared to the CM. The NSS_ven reads approximately 0.17 mg/dL lower than the CM. Based on the 95% CI, creatinine concentrations from venous aspiration samples could underestimate the true creatinine value by as much as 0.88 mg/dL or overestimate by 0.54 mg/dL as can be visualized in the Bland-Altman plot. Similarly, creatinine concentrations determined via NSS_ear could be underestimated by as much as −1.04 mg/dL or overestimated by 0.62 mg/dL as can be visualized in the Bland-Altman plot. These results further support that the NSS values, regardless of the sample type evaluated in this study, are not equivalent when compared to the CM values, possibly leading to misclassification of nonazotemic patients.

Equivalence testing further assessed the potential clinical value of the device. A constraint of 0.3 mg/dL was used such as increment aligns with the definition of acute kidney injury and serves as a prompt for intervention and enhanced monitoring.^10,11 For the equivalence test, limiting the allowed difference between tests to 0.3 mg/dL, the NSS and CM were found not equivalent. Consequently, if used interchangeably (NSS and CM), an incremental creatinine increase of ≥ 0.3 mg/dL could be missed, thereby potentially losing a critical window for early detection and treatment of acute kidney injury, making this device unsuitable for monitoring creatinine concentration trends interchangeably with the CM.

Based on our results, there is limited appropriate use of this device for screening nonazotemic pigs, and other established laboratory methods should be preferentially used. If no established device is available, or an insufficient amount of blood is available to be tested with other methods, this device, even if able to measure creatinine concentration in pigs, needs to be interpreted with extreme caution. In fact, the negative bias detected between the NSS and CM in this population suggests that a misclassification of pigs as nonazotemic is possible while mild azotemia is present. Moreover, considering the wide CI of the limits of agreement between the NSS and CM, it cannot be ruled out that a pig with normal creatinine would be classified as azotemic. Consequently, there could be added risks for the patient when administering potentially nephrotoxic drugs, or there could even be unnecessary therapeutic costs (ie, withholding of nephrotoxic medications due to a perceived risk).

The absence of azotemic patients within the enrolled population is a primary limitation of the study. A large portion of the study population was comprised of healthy animals presenting for elective procedures or pigs sampled after medical stabilization. Given the lack of azotemic pigs in the population, the NSS performance could not be comprehensively evaluated, and no definitive conclusions can be drawn on azotemic animals. The device performance on azotemia is relevant, especially when considering that, for example, in dogs it performed acceptably in normal and mildly azotemic animals but was found to be unacceptable in severely azotemic dogs and therefore not of clinical use.⁴ The only azotemic pig in this study was excluded because of being an extreme outlier; however, evaluating the performance of the device in this specific case with high creatinine the NSS gave very different values than the CM regardless of sample type. Considering the lack of equivalence in nonazotemic pigs and the large difference in this single case, the performance of the device seems unreliable in pigs, and further data collection beyond these preliminary results was not pursued.

The NSS device failed the sample analysis, resulting in an “error” notification intermittently during the study period, most often when testing NSS_ear. This may be partly attributed to the inconsistent flow of blood from an ear prick compared to the more controlled application of blood onto the reagent strip when using a syringe and needle to transfer blood from a tube. Additionally, vasoconstriction associated with hypovolemic states, fast coagulability, or anesthetic drugs may have contributed to reduced aural blood flow, resulting in poor consistency in sample collection from ear pricks. This should be a consideration when using this site for blood sampling for other devices using low blood volume samples.

In conclusion, the NSS demonstrates moderate to good internal precision. Although no definitive conclusion can be drawn on the performance of this device in azotemic pigs, in the population of nonazotemic pigs studied, the CM analyzer and the NSS results were not equivalent regardless of the sample type considered. Since this handheld device does not perform well in nonazotemic pigs, possibly misclassifying them, its clinical use for screening creatinine in pigs is not advised.

Supplementary Material

Supplementary materials are posted online at the journal website: avmajournals.avma.org.