Accuracy of Crit-Line® monitors in predicting hematocrit and change in blood volume of dogs during hemodialysis

Abstract

Background: Ultrafiltration (UF) is an extracorporeal technique for treating fluid overload and is monitored with noninvasive blood volume monitors.

Hypothesis/objectives: The primary objective was to determine the accuracy of Crit-Line® III and IV noninvasive blood volume monitors to estimate canine packed cell volume (PCV) and changes in blood volume during UF. A secondary aim was to determine accuracy of targeted ultrafiltration rates (UFR) compared to calculated delivered UFR.

Methods: An ex vivo study with a single Phoenix® X36 platform and canine packed red blood cells (pRBC) was performed. Through dilution and UF, clinically applicable PCV values were obtained and compared to Crit-Line® hematocrit (Hct). Blood volume was constantly measured and compared to targeted UFR. Systematic and proportional bias were calculated using a Bayes method.

Results: Crit-Line® III and IV reported Hct was significantly lower than PCV (n = 140, median 26%, range, 8%-50%) when PCV was >25% and >30%, respectively. Crit-Line® III and IV calculated change in blood volume (ΔBV%) was significantly different from measured blood volume changes at ΔBV% ±20% and >-20%, respectively. Comparing targeted and delivered UFR (mL/h), less than targeted UFR was removed at UFR100 through UFR400 and UFR0 adding volume.

Conclusions and clinical importance: Crit-Line® III and IV monitors provide accurate estimates of canine PCV and UF volume change within specific ranges and are useful in monitoring canine UF and hemodialysis. Veterinary hemodialysis services should evaluate individual machines for UFR inaccuracies, which can meaningfully affect small animals.

Keywords: acute kidney injury; fluid overload; intradialytic morbid event; packed cell volume; ultrafiltration.

FIGURE 1

Bias (scatter) plots of Crit‐Line® Hct and PCV measurements, shown on left y‐axis scale, versus the Bayes‐derived best linear unbiased predictor (BLUP) of the PCV in canine pRBC. The corresponding regression lines are shown for the reference standard (PCV—black solid line with y‐intercept = 0 and slope = 1) and the Crit‐Line® Hct (green dashed line). The right y‐axis scale depicts the estimated difference in the measurements, or amount of bias (shown as red dash‐dot line) versus the predicted value. Deviations of the Crit‐Line® Hct measurements from the reference line (PCV) are increasingly negative with increasing PCV. Data for Crit‐Line® III are shown in (A) and for Crit‐Line® IV in (B).

FIGURE 2

Plot showing total bias [dash‐dot line], that is, systematic + proportional, in Crit‐Line® Hct measurement as a function of the best linear unbiased predictor (BLUP) of PCV in canine pRBC, with its 95% confidence band depicted by dashed lines. Total bias for Crit‐Line® III is shown in (A) and for Crit‐Line® IV in (B). An increasing negative total bias is seen with increasing PCV in both figures. Total bias is significantly different from 0 for PCV values >25 for Crit‐Line® III and for PCV values >30 for Crit‐Line® IV.

FIGURE 3

Distribution of mild, moderate, and severe hemolysis in canine PCV and Crit‐Line® IV Hct samples, shown on Bland‐Altman style plot. Hemolysis was not significantly associated with Hct measurements in a linear mixed model (P = .11).

FIGURE 4

Bias (scatter) plots of the ΔBV% as depicted by the Crit‐Line® monitor relative to serial volume measurements and calculated change in blood volume (%) in canine pRBC. Bias plots show monitor‐reported values on left y‐axis scale, versus the Bayes‐derived best linear unbiased predictor (BLUP) of the volume change. The corresponding regression lines are shown for the reference standard (calculated volume change—black solid line with y‐intercept = 0 and slope = 1) and the Crit‐Line® monitor‐reported ΔBV% (green dashed line). The right y‐axis scale depicts the estimated difference in the measurements, or amount of bias (shown as red dash‐dot line) versus the predicted value. Deviations of the Crit‐Line® ΔBV% measurements from the reference line (calculated volume change) are increasingly negative with increasing change in blood volume. Data for Crit‐Line® III are shown in (A) and for Crit‐Line® IV in (B).

FIGURE 5

Plots showing total bias [dash‐dot line], that is, systematic + proportional, in the monitor‐depicted ΔBV% as a function of the best linear unbiased predictor (BLUP) of the calculated change in blood volume in canine pRBC, with its 95% confidence band shown by dashed lines. Total bias for Crit‐Line® III is shown in (A) and for Crit‐Line® IV in (B). Total bias was significantly different from 0 in Crit‐Line® III at larger changes in blood volume, that is, approximately ±20%, but in Crit‐Line® IV only in larger negative changes in blood volume, that is, approximately greater than −20%.

FIGURE 6

Total bias in delivered UFR relative to targeted UFR (mL/h) in a canine ex vivo study. Total bias is the constant bias (46.3 mL/h; 95% CI: 42.3‐50.3) plus proportional bias (0.977; 95% CI: 0.963‐0.991). Delivered UFR is significantly different than the targeted UFR (P < .05). Target UFR was controlled by the hemodialysis machine and delivered UFR measured by change in volume of the graduated pitcher. In this depiction, “perfect” agreement between targeted and delivered UFR would result in a line plotted at y = 0.