PEDIATRICS Vol. 106 No. 3 September 2000, pp. 497-504

Clinical Performance of an In-Line Point-of-Care Monitor in Neonates

John A. Widness, MD^*, Jeff C. Kulhavy, MTASCP, Karen J. Johnson, RN^*, Gretchen A. Cress, RN^*, Irma J. Kromer, BA^*, Michael J. Acarregui, MD^*, and Ronald D. Feld, PhD

From the ^* Department of Pediatrics, Children's Hospital of Iowa, and the  Department of Pathology, University of Iowa, Iowa City, Iowa.

	ABSTRACT

Top Abstract MaterialsMethods Results Discussion Conclusion References

Objective.  To evaluate the bias, precision, and blood loss characteristics of an ex vivo in-line point-of-care testing blood gas and electrolyte monitor designed for use in critically ill newborn infants.

Study Design.  Study participants included consecutive neonates with an umbilical artery catheter (UAC) in use for clinical laboratory testing. The in-line monitor (VIA LVM Blood Gas and Chemistry Monitoring System, VIA Medical, San Diego, CA) was directly connected to the participant's UAC and the monitor's determinations of pH, PCO₂, PO₂, sodium, potassium, and hematocrit (Hct) were compared with those simultaneously drawn and measured with a standard bench top laboratory instrument (Radiometer 625 ABL; Radiometer America, Inc, Westlake, OH). The bias (the mean difference from the reference method) and precision (1 standard deviation of the mean difference) performance criteria of the in-line monitor were derived using standard laboratory procedures.

Results.  Sixteen neonates monitored for a total of 37 days had a total of 229 paired blood samples available for comparison by the 2 methods. Bias and precision performance characteristics of the in-line monitor were similar to reports of other point-of-care devices (ie, pH: .003 ± .024; PCO₂: .35 ± 2.84 mm Hg; PO₂: .39 ± 7.30 mm Hg; sodium: .52 ± 2.34 mmol/L; potassium: .17 ± .18 mmol/L; and Hct: .61 ± 2.80%). The range of values observed for each parameter included much of the range anticipated among critically ill neonates (ie, pH: 7.15-7.65; PCO₂: 25-75 mm Hg; PO₂: 25-275 mm Hg; sodium: 127-150 mmol/L; potassium: 2.6-5.5 mmol/L; and Hct: 32%-60%). Mean blood loss (± standard deviation) per sample with the in-line monitor was approximately one-tenth that of the reference method: 24 ± 7 µL versus 250 µL, respectively. There was no evidence of hemolysis and no patient related safety issues were identified with use of the in-line monitor.

Conclusions.  Repeated laboratory testing of critically ill neonates using an ex vivo in-line monitor designed for use in neonates provides reliable laboratory results. The blood loss and hemolysis data obtained suggests that this monitoring device offers potential for reducing neonatal blood lossand possibly transfusion needsduring the first weeks of life. Before this promising technology can be routinely recommended for care of critically ill neonates, greater practical experience in a variety of clinical settings is needed.  Key words:  in-line point-of-care testing, phlebotomy, anemia.

Frequent laboratory phlebotomy loss leads to anemia among critically ill low birth weight infants.^1-3 Recent improvements in bedside point-of-care near patient analyzers and in in-line in vivo and ex vivo monitors offer the possibility of reducing blood lossand therefore of preventing anemia and reducing blood transfusionsamong this patient group.⁴ Compared with bench top analyzers, point-of-care devices offer additional advantages in providing immediate results without the need to transport samples to the laboratory while reducing the chance of preanalytic error. Relative to adults,^5-20 there are few reports on the performance characteristics of point-of-care devices in infants and children.^21-23

An in-line ex vivo monitor approved by the Food and Drug Administration (FDA) and capable of measuring pH, blood gases, electrolytes, and hematocrit (Hct) in adults²⁰ has recently been adapted for use in premature low birth weight infants and other volume-restricted patients. The purpose of the present study was to evaluate the bias, precision, and blood loss characteristics of this monitor for neonates in the clinical setting. We hypothesized that compared with bench top analyzers used in our laboratory (and to point-of-care devices reported in the literature), the in-line ex vivo neonatal monitor would perform in an equivalent mannerbut with negligible blood loss.

	MATERIALS AND METHODS

Top Abstract MaterialsMethods Results Discussion Conclusion References

This study was approved by the local institutional review board. Before study, informed written consent was obtained from 1 or both parents.

Entry Criteria

Eligible infant study participants included those admitted to our neonatal intensive care unit (NICU) who had an umbilical artery catheter (UAC) catheter already in use for blood sampling and blood pressure monitoring. Because of the developmental nature of the present study and because of the possibility of inadvertent blood loss as a result of instrument or operator error, enrollment was restricted to infants with birth weights >1 kg.

Study Design

Consecutive blood laboratory samples ordered for clinical purposes were simultaneously analyzed both by a reference instrument (Radiometer Model ABL625; Radiometer America, Inc, Westlake, OH) and the in-line point-of-care test device (VIA LVM Blood Gas and Chemistry Monitoring System, VIA Medical, San Diego, CA) designed for neonates and other volume-restricted patients. Blood samples analyzed by the reference method were drawn from a stopcock located between the patient and the in-line sensor. Reference samples were drawn in the interval during which the monitor analysis was taking place. Analysis of reference samples was performed within 5 minutes of sampling in the satellite laboratory adjacent to the NICU.

Operation and Features of the Point-of-Care Monitor Device

The VIA LVM Blood Gas and Chemistry Monitoring System is an FDA-approved, Class II, Clinical Laboratories Improvement Act (CLIA)-exempt medical device consisting of a monitor and associated disposables that are directly connected to the patient's intravascular catheter (Fig 1). Like its adult predecessor, the neonatal monitor uses microelectrochemical detection for measuring pH, PO₂, PCO₂, potassium, sodium, and Hct.²⁰ The low-volume neonatal system is designed to restrict the amount of flush administered. The disposables include: 1) a 500-mL bag of isotonic calibration fluid (Normosol-R, Abbott Laboratories, Chicago, IL, containing 140 mmol/L sodium, 5 mmol/L potassium, 3 mmol/L magnesium, 98 mmol/L chloride, 27 mmol/L acetate and 23 mmol/L gluconate) to which 10 mmol bicarbonate and 500 U heparin are added; 2) a sensor heated to 37°C for the measurement of the 6 study parameters; 3) a proprietary dial stopcock for directing the flow of the parenteral fluids and blood samples; and 4) a collection bag for the disposal of calibration fluid and blood lost during analysis. To improve accuracy and reduce blood loss, the neonatal monitor incorporates small diameter tubing, tubing connectors that reduce line turbulence and an optional presample flush of the patient's UAC fluid equivalent to the combined dead space of the line and UAC.

View larger version (81K): [in this window] [in a new window]   Fig. 1.   Diagram of the VIA LVM in-line ex vivo monitor and its components at the bedside when in operation and attached to the patient's UAC.

Operation of the in-line monitor was performed as recommended by the manufacturer. Before connection to the patient, each sensor undergoes an initial 2-point calibration. The minimum interval between blood sampling for the in-line monitor was 12 minutes. While connected to the patient in stand-by mode, calibration fluid is slowly (ie, 5 mL/h), but continuously, infused through the sensor tubing and into the collection bag. A 1-point calibration is performed automatically every 30-minutesor sooner, if directed by the operator. The calibration fluid is changed daily while the sensor, stopcock, and collection bag are changed as frequently as mandated by hospital regulations for parenteral fluid tubing changes. With the exception of these tubing changes, the operation of the in-line monitor takes place in a closed environment, thus avoiding exposing medical personnel to the patient's blood while at the same time reducing the number of times that the arterial catheter line is entered.

Operator initiation of in-line blood analysis with the monitor necessitates brief interruption of monitoring and parenteral fluid administration through the catheter. The withdrawal of 1.5 mL of blood into the sensing chamber takes 54 seconds. Sample analysis requires an additional 70 seconds and allows time for a paired UAC blood sample to be drawn from a proximal in-line stopcock. Following analysis, blood is returned via the catheter along with .5 mL of calibration solution to reduce blood loss. Results displayed on the front panel of the monitor are also printed out.

Determination of Blood Volume Loss and Hemolysis Associated With Monitor Use

The volume of blood lost during sampling was determined after the performance of 7 or more sequential blood analyses in which sensor fluid remaining after the .5-mL postsample flush was discharged into the same collection bag. The volume of blood lost per analysis was calculated based on the number of analyses performed, the volume of red blood cells (RBCs) present in the collection bag, and the patient's average Hct during the sampling period.

To determine the extent of hemolysis associated with blood sampling, bench top experiments were performed using fresh (<24-hour-old) heparinized adult and fetal umbilical cord blood sampled through a 3.5 Fr umbilical arterial catheter (Kendall-LTP, Chicopee, MA). Operation of the low-volume monitor was identical to that described above with the exception that instead of returning the blood sample to the patient (ie, to the bag containing heparinized fetal or adult blood), the sample was discharged into a test tube through the in-line stopcock port. This blood was then centrifuged and plasma hemoglobin determined spectrophotometrically with the Hitachi 747 (Boehringer Mannheim Corporation, Indianapolis, IN).

Statistical Analysis

Data are presented as the mean ± standard deviation. An  value <.05 was considered statistically significant. Bias and precisionthe study's primary outcome variableswere determined by the method of Bland and Altman.²⁴ Using this method, bias (an indicator of accuracy) is defined as the mean difference of the results of the in-line monitor and the reference method; and precision (an indicator of intersample reproducibility) is defined as 1 standard deviation of the differences between the 2 methods. Data were also subjected to an outlier rejection procedure routinely used by clinical laboratories in eliminating inexplicable, markedly discrepant results.²⁵ In this latter procedure, outlier results are defined data points in which the difference between the 2 methods exceeds 4 times the absolute mean difference for all of the data points. Up to 2.5% of data points may be excluded. Agreement of the 2 methods was also assessed by determination of Pearson correlation coefficients for each laboratory parameter.

	RESULTS

Top Abstract MaterialsMethods Results Discussion Conclusion References

Gestational age and weight at birth for the 12 male and 4 female study participants were 36.8 ± 4.8 weeks and 2743 ± 1046 g, respectively. All but 2 participants were outborn. Point-of-care monitoring was begun at 2.5 ± 2.9 days postnatal age and continued for an average of 2.3 ± .7 days. Clinical diagnoses at the time of study entry included primary pulmonary hypertension (n = 6), congenital heart disease (n = 3), respiratory distress syndrome (n = 3), pneumonia (n = 2), sepsis (n = 1), and hypoxic ischemic encephalopathy (n = 1).

Analysis of Paired Blood Samples

Among the 16 study participants there were 229 paired blood sample results deemed valid for inclusion in the analysis. All were ordered for clinical purposes. An additional 95 paired samples were excluded because of an incorrect presample flush volume. In 49 of these an incorrect presample volume was administered while in the remaining 46 there was an operator error. The occurrence of events leading to samples excluded for these reasons was evenly distributed throughout the study period. Both errors were later corrected by software modifications. Two incapacitating monitor malfunctions (crashes) occurred as a result of another software error. This problem was also corrected. Only 1 patient experienced an acute clinical deterioration event necessitating laboratory testing more frequently than the 12 minutes interval allowed by the in-line monitor at the time of study. (This interval currently is 6 minutes). Finally, 2 data pairs were eliminated because of unexplained operator errors, probably related to the entry of incorrect monitor instructions.

Although results of all 229 valid paired analyses were available for pH, PCO₂ and PO₂, reference laboratory method determinations were not always available for sodium, potassium, and Hct. These discrepancies were overwhelmingly attributable to the reference laboratory not analyzing all 6 study parameters as a result of the patient's physician not having ordered them. This included 1 determination each for pH, PO₂, PCO₂, and Hct; 46 for sodium, and 47 for potassium. Of the remaining data pairs, 1.2% of the data points for individual study parameters were omitted from the final analyses because they met the outlier criteria. These outliers included 5 sample pairs for Hct, 3 each for PCO₂, PO₂, and sodium, and 1 for potassium. Among the outliers, 4 sample pairs were identified in which 2 or 3 of the study parameters in the same blood sample were identified as outliers. These included 4 Hct, 3 PCO₂, and 2 sodium data pairs. In all 4 instances, results of samples drawn immediately before and immediately after the outlier data points more closely approximated the in-line monitor's values than they did those of the reference method. The observation that the reference instrument's values were all lower in these few outlier data points suggested that the reference sample was most likely diluted with residual flush solution inadvertently left in the stopcock.

There were no patient-related injuries identified with the use of the in-line monitor. Although an occasional defective sensor was found during initial calibration (but before being placed on the patient), there were no episodes of sensor malfunction while in use on patients. Among the 324 blood samples analyzed, there were 3 occasions when operator error resulted in the inadvertent flushing of the participant's entire 1.5-mL blood sample into the collection bag. Consistent with the monitor's software design, there were no instances of excessive fluid boluses being inadvertently administered to patients.

All 6 study parameters showed good agreement in the correlation observed between the reference and the in-line monitor results (Fig 2). Correlation coefficients for all 6 were highly significant with all r values >.95, except sodium and Hct, which were .90 and .89, respectively. The range of values encountered for each of the 6 laboratory parameters included those commonly encountered among critically ill neonates (pH: 7.15-7.65; PCO₂: 25-75 mm Hg; PO₂: 25-275 mm Hg; sodium: 127-150 mmol/L; potassium: 2.6-5.5 mmol/L; and Hct: 32%-60%).

View larger version (42K): [in this window] [in a new window]   Fig. 2.   Correlation of the reference method with the in-line monitor results. The line of identity is indicated by the diagonal dashed line and the solid line representing the best-fit regression line for the range of values included. Data indicating the number of points included in the analysis, the correlation coefficient, the formula for the regression line is included with the individual plots.

Bland-Altman plots of the 6 study parameters revealed biases that were all close to 0, and thus of no clinical significance (Fig 3). The vertical scatter of the paired data points about the bias in 5 of the 6 Bland-Altman plots fell within limits used to define acceptable CLIA performance criteria, ie, 80% of the values fell within the CLIA-defined limits. Only Hct fell outside of CLIA performance criteria.

View larger version (46K): [in this window] [in a new window]   Fig. 3.   Bland-Altman plots for the 6 study parameters, ie, mean value of the reference and the in-line monitor determinations versus the difference between the reference and the VIA methods. The thick solid horizontal line indicates the bias while the 2 parallel dashed lines indicate the CLIA performance range (ie, 80% of the VIA values are expected to fall between the 2 dashed lines) referenced to the monitor's bias.20 The percentage of outliers in each panel indicates the total number of paired samples that fell outside of the CLIA range.

With the exception of pH, analysis of bias and precision results for the 95 paired samples in which the wrong presample flush volume was administered were statistically indistinguishable from those obtained for the correct presample flush volume (data not shown). In the case of pH, bias and precision were moderately increased compared with correct presample flush volume data, ie, from .003 to .011 and from .024 to .029 pH units, respectively. This was almost exclusively attributable to 6 pH values that would have been excluded as statistical outliers in the analysis of the other 228 paired samples.

Comparison of VIA Monitor's Bias and Precision to Other Point-of-Care Devices

When the bias and precision results of the 6 analytes included in the present study were compared with results taken from the literature for the 3 general categories of point-of-care devices, ie, other in-line ex vivo monitors,^14-19 in-line in vivo monitors,^12,13,21,23 and near patient analyzers,^5-11,22 the results were similar (Fig 4). Bias (depicted by the solid horizontal bars in Fig 4) differences reported for all 3 categories of devices were small and clinically insignificant. Precision (depicted by the ± 1 standard deviation error bar in Fig 4), was worse for PO₂ determined using in-line in vivo monitors than with the other monitors. This is likely related to the well-known intrinsic perturbation of PO₂ readings resulting from the movement of optode sensors in close proximity to blood vessel walls.²¹ Finally, VIA monitor's precision results for sodium, potassium and Hct using the paired sample technique was not as precise as those obtained with near patient analyzers using the split sample technique.

View larger version (32K): [in this window] [in a new window]   Fig. 4.   Comparison of mean bias and precision for the 3 different categories of point-of-care devices taken from the literature, ie, in-line in vivo monitors (eg, NeoTrend, Diametrics Medical, St Paul, MN),12,13,21,23 other in-line ex vivo monitors (none commercially available at present)14-19 and near-patient analyzers (eg, i-STAT, Princeton, NJ)5-11,22 with these same data from the present study using the in-line ex vivo VIA LVM monitor adapted for neonates. Bias is indicated with the horizontal solid bars and precision indicated with the thin horizontal error bar.

Blood Loss and Hemolysis With VIA Monitor

Blood loss per sample analyzed by the in-line monitor was 24 ± 7 µL (n = 24 determinations)an order of magnitude less than the minimum 250 µL volume of blood requested for testing by the reference method. In testing for hemolysis, no difference was identified in the concentration of plasma hemoglobin measured before and after analysis in either adult or fetal umbilical cord blood. In all samples plasma hemoglobin concentration was <40 mg/dL.

	DISCUSSION

Top Abstract MaterialsMethods Results Discussion Conclusion References

In the present study, bias and precision for pH, PCO₂, PO₂, sodium, and potassium measured with an in-line ex vivo point-of-care monitor adapted for use in neonates exceeded laboratory testing performance criteria established by CLIA.²⁶ Although similar precision and bias performance levels have been previously achieved among adult volunteers using a similar monitoring device,²⁰ important design changes were required before comparable results were achievable with the neonatal version. Importantly, blood loss associated with the use of this modified device was an order of magnitude less than that with conventional laboratory testing. There was no evidence for hemolysis resulting from the use of the modified monitor. Because laboratory testingand hence iatrogenic blood lossis greatest in the first weeks of life when infants are most critically ill, use of the neonatal low-volume monitor has the potential for slowing the development of anemia. This, in turn, could result in a modest reduction in the number of RBC transfusions these infants receive while providing essential laboratory data in a reliable manner and with a short turnaround time.

Bias and Precision

Because of differences in the clinical management of critically ill neonates compared with adults, modifications in the design and operation of the comparable monitor used in adults were necessary. The primary differences between the 2 patient groups are that infants commonly receive a broader spectrum of parenteral fluids through catheters than adults and that infants (and other volume-sensitive patients) would rapidly become fluid overloaded if they received the 8 mL saline bolus flush used in returning blood to adults. Modifications incorporated into the neonatal in-line device to overcome these problems included using smaller bore tubing, reducing line turbulence at tubing connections and including a customized presample flush volume based on the dead space of the monitor's line and attached catheter. As demonstrated with the bias and precision data of the present study, laboratory resultswith the exception of Hctachieved with the modified neonatal in-line monitor exceed the CLIA performance criteria and are equivalent to those reported for other point-of-care analyzers and monitors. The less than desirable precision results obtained for Hct could be attributable to differences in the methodologies used by the 2 instruments, ie, the in-line monitor measures Hct by conductance while the reference instrument determines hemoglobin more precisely using spectrophotometric methodology.

Although similar bias and precision results were found for all but pH when an incorrect presample flush volume was used, the large number of pH outliers encountered when an incorrect presample flush volume was administered suggests that the presample flush is important in avoiding spurious results. Whether similar perturbations in performance will occur in association with other parenteral solutions, eg, those containing protein or amino acids, when insufficient flush volumes are administered is uncertain based on our limited data, eg, only 9 samples were drawn when participants were receiving amino acid containing fluids. However, based on the present study's bias and precision results compared with those of other point-of-care devices, we believe that the in-line system tested is sufficiently robust that spurious laboratory results are highly unlikely when the correct volume for presample flush is administered.

In bias and precision comparisons of patient sample results, in-line monitoring devices are at a disadvantage relative to near patient point-of-care analyzers. For the latter, bias and precision are based on split sample results. In contrast, in-line monitor bias and precision determinations of patient samples typically relies on comparisons of the results of separate blood samples drawn several seconds to a minute apart. While technically feasible to perform split sample analysis using the in-line monitor, doing so is cumbersome and it does not allow blood to be returned to the patient in a sterile manner.

Technical Problems Encountered With Monitor

Although the determination of bias and precision of the VIA LVM in-line monitoring system and comparison of these data with reports of other studies in the literature were our primary focus, the study was done while the in-line monitor was undergoing significant software modifications. Mostbut not allof the problems we encountered were of a software nature and most have been corrected or improved. As the study progressed, it was evident that these modifications resulted in improvements that made the monitor easier to use. Problems encountered and later resolved included: 1) the wrong presample flush volume being given when the monitor was powered on and off; 2) lack of clarity of messages appearing on the monitor screen; 3) gas bubbles forming in the sensing device and tubing; 4) inability of the operator to manually initiate a 1-point calibration on command and print out these results; 5) the interval between repeated samples was excessive; and 6) excessive time in obtaining results if the monitor was undergoing an automated 1-point calibration. For the last 2 of these, the time interval has been decreased from 12 to 6 minutes. It is likely that future software and hardware modifications will lead to further improvements in ease of use. As future clinical needs evolve in the NICU, it seems likely that new software and hardware will need to evolve as well. Examples of these future clinical challenges may include effectively interfacing with multilumen catheters and including additional analytes (eg, glucose, creatinine, blood urea nitrogen, lactate, calcium, chloride, and bilirubin) within the sensing unit without additional blood loss.

Despite the robust performance exhibited by the present monitor, several problems exist with the current configuration of the low-volume in-line system in the present study. Although some of these are amenable to correction, others may not be. The occasional occurrence of operator error resulting in inadvertent flushing of the 1.5-mL blood sample into the collection bag can be reduced through software modifications and through better education to ensure greater familiarity and expertise on the part of users. It is uncertain whether blood loss and the volume of fluid boluses needed for returning blood can be further reduced. Before every blood sample withdrawn by the monitor, approximately .2 to .4 mL of the parenteral fluid is first infused through the catheter line (depending on the size of the umbilical catheter). This is in addition to the .5-mL postsample flush of calibration fluid containing .07 mmol sodium, .003 mmol potassium, .05 mmol chloride, and .5 U heparin. Just as with conventional blood sampling, if monitor sampling is done too often, there is the potential for administering excessive amounts of fluids and electrolytesparticularly among very low birth weight (VLBW) infants whose blood volumes are limited. However, it is precisely this group that may benefit the most from the monitor's ability to reduce blood loss.

Comparison of Bias and Precision Results With Other In-Line Monitors

Compared with standard laboratory testing instrumentation, point-of-care bedside analyzers and monitors offer advantages in making commonly used laboratory results immediately available, potentially with less blood loss, and without the need to transport samples to the laboratory. Which testing devices are best suited for which applications remain unresolved. In the NICU environment, in-line ex vivo monitoring systemssuch as the one used in the current studyoffer several advantages over other point-of-care devices. These include: 1) lower blood loss and a reduction in medical staff exposure to blood compared with near patient analyzers (but not in-line in vivo monitors); 2) being capable of determining a larger number of analytes than is possible with current in vivo monitors that can only determine pH, PCO₂ and PO₂; 3) ease of use and time savings in not having to pass the wire electrode sensor through the catheter to reside in the vascular lumen as is necessary with in vivo monitors; and 4) no chance of additional endothelial damage with the sensor tip extruding from the end of the catheter, no possibility for the sensor wire obstructing the catheter to result in inaccurate in-line pressure tracings, and no risk of damage to the electrode wire resulting from kinking within the catheter as is possible with in-line in vivo monitors.

Blood Loss During the Neonatal Period

Iatrogenic blood loss has a considerable impact on the number of RBC transfusions received by VLBW infants in the first critical weeks of life. In-line monitoringsuch as that used in the present studyoffers potential for reducing the relatively large number of transfusions these infants typically receive.²⁷ Approximately half of all RBC transfusions received by VLBW infants are administered in the first 2 weeks of lifea time when the volumes of blood transfused approximate those removed for laboratory testing.^1-3 In the present study hemolysis was not detected and blood loss was <25 µL per sampleapproximately one-tenth the 250 µL required for analysis by bench top instruments and approximately half the volume required by near patient analyzers. The amount of blood lost with in-line ex vivo monitoring is equivalent to hidden blood loss associated with laboratory testing as a result of blood remaining in tubing and sampling syringes and on gauze pads, bandages, and bedding. While detailed quantification of the reduction in blood loss attributable to laboratory testing awaits the results of future clinical studies, phlebotomy data from a cohort of 10 of our NICU infants whose birth weights were <1.2 kg suggests that a 50% reduction in laboratory blood loss is possible for critically ill infants having repeated laboratory tests drawn from UAC catheters. Currently 50% to 60% of such infants in our NICU are managed with UAC catheters, with the mean duration of catheter use being 6.1 ± 4.0 days. Gravimetrically determined weekly blood loss among these 10 critically ill infants we surveyed was 38 ± 24 mL/kg and the volume of blood transfused by 2 weeks of age was 42 ± 10 mL/kg. If in the future, additional analytes are incorporated into the ex vivo sensor, our estimate of the decrease in laboratory blood loss based on these 10 infants approached 70% to 80%.

Cost of the In-Line Monitoring System

Decisions regarding the cost-effectiveness of this device have yet to be reported. The present charge for a sensor is approximately $300with the accompanying monitor usually provided at no additional charge. Important considerations in the cost-effectiveness of in-line monitors include the cost of laboratory supplies, equipment, and personnel and the duration that the sensor is permitted to remain in use before being replaced as dictated by hospital infection control policies. The opportunity costs of point-of-care testing and the potential for low costs resulting from changes in hospital-wide delivery of laboratory services are also worthy of consideration.²⁸ If RBC transfusions decrease as a result of reductions in phlebotomy blood loss, this too will have an impact on cost.

	CONCLUSION

Top Abstract MaterialsMethods Results Discussion Conclusion References

In summary, laboratory monitoring of critically ill neonates using in-line ex vivo devices adapted for use in neonates offers advantages over other types of point-of-care testing devices. Advantages of in-line ex vivo monitors include their ability to substantially reduce blood loss while simultaneously providing frequent, reliable laboratory results with rapid turnaround time. The number of parameters measured with in-line ex vivo devices using electrochemical detection is large relative to other in-line monitors and the use of optodes. Although the cost-effectiveness of these devices is uncertain, cost considerations related to the potential of these devices to indirectly improve clinical outcomes, eg, by a reduction in the need for blood transfusion, should be considered. Greater experience with the in-line ex vivo monitors in a large number of nursery settings will permit a better assessment of the practical utility of such instruments. Before the use of these devices is allowed to become standard of care, it is important that much of this experience be subjected to carefully performed clinical trials and other hypothesis-based studies weighing the evidence for or against implementing this new, but promising, technology.^4,29

	ACKNOWLEDGMENTS

This study was supported in part by the National Institutes of Health General Clinical Research Centers Program Grant RR00059, the Children's Miracle Network, and by the VIA Medical Corporation, San Diego, California.

We thank Leon F. Burmeister, PhD, for his statistical advice, Robert L. Schmidt for his technical expertise, Mark A. Hart for his secretarial assistance, the NICU nursing and laboratory staffs for their help and support, and the families of study participants for their participation.

	FOOTNOTES

Dr Widness is an unpaid member of VIA Medical Corporation's Scientific Advisory Board. None of the authors holds any financial interest in VIA Medical Corporation.

Received for publication Jun 21, 1999; accepted Dec 22, 1999.

Reprint requests to (J.A.W.) University of Iowa Hospitals and Clinics, Department of Pediatrics, W222-1 GH, 200 Hawkins Dr, Iowa City, IA 52242-1083. E-mail: john-widness{at}uiowa.edu

	ABBREVIATIONS

FDA, Food and Drug Administration; Hct, hematocrit; NICU, neonatal intensive care unit; UAC, umbilical artery catheter; CLIA, Clinical Laboratories Improvement Act; RBC, red blood cell; VLBW, very low birth weight.

	REFERENCES

Top Abstract MaterialsMethods Results Discussion Conclusion References

1. Blanchette V, Zipursky A Assessment of anemia in newborn infants. Clin Perinatol. 1984; 11:489-516 [Medline]
2. Obladen M, Sachsenweger M, Stahnke M Blood sampling in very low birth weight infants receiving different levels of intensive care. Eur J Pediatr. 1988; 147:399-404 [CrossRef][Medline]
3. Ringer SA, Richardson DK, Sacher RA, Keszler M, Churchill WH Variations in transfusion practice in neonatal intensive care. Pediatrics. 1998; 101:194-200 [Abstract/Free Full Text]
4. Kost GJ, Ehrmeyer SS, Chernow B, The laboratory-clinical interface: point-of-care blood testing. Chest. 1999; 115:1140-1154 [Abstract/Free Full Text]
5. Gault MH, Harding CE Evaluation of i-STAT protable clinical analyzer in a hemodialysis unit. Clin Biochem. 1996; 29:117-124 [CrossRef][Medline]
6. Jacobs E, Vadasdi E, Sarkozi L, Coman N Analytic evaluation of i-STAT protable clinical analyzer and use by nonlaboratory health-care professionals. Clin Chem. 1993; 39:1069-1074 [Abstract/Free Full Text]
7. Shelledy DC, Smith WA A comparison of the i-STAT system and Corning 278 for the measurement of arterial blood gas values. Respir Care 1997; 42:693-697
8. Lindemans J, Hoefkens P, VanKessel AL, Bonnay M, Külpmann WR, VanSuijlen JEE Portable blood gas and electrolyte analyzer evaluated in a multiinstitutional study. Clin Chem. 1999; 45:111-117 [Abstract/Free Full Text]
9. MacIntyre NR, Lawlor BL, Carstens D, Yetsko D Accuracy and precision of a point-of-care blood gas analyzer incorporating optodes. Respir Care. 1996; 41:800-804
10. Wahr JA, Lau W, Tremper KK, Hallock L, Smith K Accuracy of a new, portable, hand-held blood gas analyzer, the IRMA. J Clin Monit. 1996; 12:317-324 [CrossRef][Medline]
11. Zaloga GP, Roberts PR, Black K Hand-held blood gas analyzer is accurate in the critical care setting. Crit Care Med 1996; 24:957-962 [CrossRef][Medline]
12. Zollinger A, Spahn DR, Singer T, Accuracy and clinical performance of a continuous intra-arterial blood-gas monitoring system during thoracoscopic surgery. Br J Anesthesiol. 1997; 79:47-52
13. Abraham E, Gallagher TJ, Fink S Clinical evaluation of a multiparameter intra-arterial blood-gas sensor. Intensive Care Med. 1996; 22:507-513 [Medline]
14. Clifford PS, Muzi M Validity of a patient-dedicated blood gas monitoring system. Am J Respir Crit Care. 1996; 153:A602
15. McKinley BA, Parmley CL Preliminary clinical trial of an ex vivo arterial blood gas monitor. J Crit Care. 1997; 12:214-220 [CrossRef][Medline]
16. Myklejord DJ, Pritzker MR, Nicoloff DM, Emery AM, Emery RW Clinical evaluation of the on-line Sensicathô blood gas monitoring system. The Heart Surgery Forum #1998-12101. 1998; 1:60-64
17. Mahutte CK On-line arterial blood gas analysis with optodes: current status. Clin Biochem 1998; 31:119-130 [CrossRef][Medline]
18. Haller M, Kilger E, Briegel J, Forst H, Peter K Continuous intra-arterial blood gas and pH monitoring in critically ill patients with severe respiratory failure: a prospective, criterion standard study. Crit Care Med. 1993; 22:580-586
19. Shapiro BA, Mahutte CK, Cane RD, Gilmour IJ Clinical performance of a blood gas monitor: a prospective multicenter trial. Crit Care Med. 1993; 21:497-494
20. Bailey PL, McJames S, Cluff ML, Wells DT, Orr JA, Wenternskow DR Evaluation in volunteers of the VIA V-ABG automated bedside blood gas, chemistry, and hematocrit monitor. J Clin Monit Comput. 1998; 14:339-346 [CrossRef][Medline]
21. Morgan C, Newell SJ, Ducker DA, Continuous neonatal blood gas monitoring using a multiparameter intra-arterial sensor. Arch Dis Child. 1999; 80:F93-F98
22. Murthy JN, Hicks JM, Soldin SJ Evaluation of i-STAT portable clinical analyzer in a neonatal and pediatric intensive care unit. Clin Biochem. 1997; 30:385-389 [CrossRef][Medline]
23. Weiss IK, Fink S, Harrison R, Feldman JD, Brill JE Clinical use of continuous arterial blood gas monitoring in the pediatric intensive care unit. Pediatrics. 1999; 103:440-445 [Abstract/Free Full Text]
24. Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet. 1986;307-310
25. National Committee for Clinical Laboratory Standards. Method Comparison and Bias Estimation Using Patient Samples (NCCLS Document EP09-A). National Committee for Clinical Laboratory Standards; 1995
26. Ehrmeyer SS, Laessig RH, Leinweber JE, Oryall JJ Medicare/CLIA final rules for proficiency testing: minimum intralaboratory performance characteristics (CV and bias) needed to pass. Clin Chem. 1990; 36:1736-1740 [Abstract/Free Full Text]
27. Widness JA, Seward VJ, Kromer IJ, Burmeister LF, Bell EF, Strauss RG Changing patterns of red blood cell transfusion in very low birth weight infants. J Pediatr. 1996; 129:680-687 [CrossRef][Medline]
28. Kilgore ML, Steindel SJ, Smith JA Cost analysis for decision support: the case of comparing centralized versus distributed methods for blood gas testing. J Healthcare Manage. 1999; 44:207-215 [Medline]
29. Rainey PM Outcomes assessment for point-of-care testing. Clin Chem. 1998; 44:1595-1596 [Free Full Text]