The Rise and Fall of i-stat Point-of-Care Blood Gas Testing in an Acute Care Hospital

Transcription

1 Clinical Chemistry / POINT-OF-CARE BLOOD GAS TESTING The Rise and Fall of i-stat Point-of-Care Blood Gas Testing in an Acute Care Hospital Valerie L. Ng, MD, PhD, 1,2 Roger Kraemer, CRT, 3 Carol Hogan, MT(ASCP), MBA, 2 * Doug Eckman, RRT, MBA, 3 and Mark Siobal, BS, RRT 3 Key Words: Acute care hospital; Point of care; ABGs; Blood gas; Ionized calcium; i-stat; Quality control; Cost savings Abstract In response to a $35, laboratory budget cut and closure of an intensive care unit based laboratory and a desire to maintain turnaround times of 1 minutes or less, a multidisciplinary group developed and implemented point-of-care (POC) testing. Only blood gases (ph, PO 2, and PCO 2 ) and ionized calcium values were deemed essential stat tests. Three commercially available POC blood gas devices were evaluated; all yielded results comparable to in-house reference methods. The 1 device with a US Food and Drug Administration approved method for ionized calcium testing and with an existing interface for laboratory information systems was selected. Fiscal analysis predicted annual savings of approximately $225,. POC blood gas analysis was implemented in April 1996 coincident with closure of the intensive care unit based laboratory. Clinical laboratories and POC blood gas test volumes remained constant through August 1998; in contrast, the number of ionized calcium tests decreased dramatically after April In August 1998, clinically significant (ie, artificial ventilation parameters would have been altered based on test results) discrepant PCO 2 values were observed sporadically and noted only with patient specimens, not with commercial controls or electronic simulators. Because investigation failed to identify the cause, use of the POC device was discontinued in September The explosive growth of managed health care plans in the early 199s created tremendous pressures on health care delivery systems to provide economic and efficient patient care. The shift of patient care to outpatient health care facilities for management or prevention of chronic diseases resulted in an excess of hospital beds and underutilization of hospital-based services. For patients requiring hospitalization, there has been an increasing trend to minimize length of stay while maximizing the level of service provided. Hospital-based services thus have been placed in a precarious position their very survival is in question given the excess number of hospital beds in the marketplace, yet they are pressured to provide more intensive inpatient services to minimize lengths of hospitalization. 1 San Francisco General Hospital (SFGH), San Francisco, CA, is an acute care hospital that provides a disproportionate amount of medical care to the medically indigent and MediCal population. During the early 199s, given increased costs of health care and decreased reimbursements, operating budgets for all clinical departments at SFGH underwent intense scrutiny and cuts. In 1995, the SFGH Clinical Laboratories budget was reduced by $35,, with the expectation that the level of service would be maintained. Evaluation of existing operations revealed duplication of laboratory testing between a dedicated laboratory serving only the intensive care units (ICU laboratory) and the main clinical laboratories. Duplicated tests included blood gas analysis, co-oximetry, limited electrolyte analysis of whole blood (eg, Na+, K+, and ionized calcium levels), and lactate determinations. The average turnaround time was 1 minutes or less for specimens analyzed in the ICU laboratory. An understaffed messenger service and a pneumatic 128 Am J Clin Pathol 2;114: American Society of Clinical Pathologists on 3 January 218

2 Clinical Chemistry / ORIGINAL ARTICLE tube system installed during the early 197s that lacked tracking or soft-landing capability precluded reliable rapid transport of specimens and centralization of the duplicated testing. Point-of-care (POC) test devices for these tests recently had become commercially available. It was recognized that considerable savings could be achieved if POC testing could be absorbed by nonlaboratory personnel, permitting closure of the ICU laboratory and elimination of the costs associated with its staffing. In this article, we detail the medical and financial issues considered before implementation of POC blood gas testing and report our experience with this POC program during the ensuing 3 years. Phase 1: Environment, Assessment, and Planning SFGH and the SFGH Clinical Laboratories SFGH is a 275-bed licensed acute care hospital owned by the city and county of San Francisco and operated under the auspices of the San Francisco Department of Public Health. The Clinical Laboratories are operated under the auspices of the Department of Laboratory Medicine at the University of California San Francisco on a contractual basis with the City and County of San Francisco. Testing and personnel requirements for the SFGH Clinical Laboratories License is regulated by the Clinical Laboratory Improvement Amendments of 1988 (CLIA 88), the Joint Commission on Accreditation of Healthcare Organizations, and the Business and Professions Code (Division 2, Chapter 3) of the State of California. The ICU laboratory was established in the mid-197s and was administered by the SFGH Anesthesia and Peri-Operative Services. The ICU laboratory existed to provide rapid service to 3 adult ICUs and the operating room. In 1993, administrative oversight of the ICU laboratory was transferred to the Clinical Laboratories. When responsibility was transferred, the ICU laboratory had a full-time medical director, was open 24 hours daily, and was staffed by 5 licensed clinical laboratory technologists. The ICU laboratory offered arterial blood gas analysis, co-oximetry, limited whole blood electrolyte analysis (ie, K+, Na+, ionized calcium), and lactate level determinations. In addition and related to their instrumentation, an ionized calcium level was determined with every blood gas analysis. The entire test menu of the ICU laboratory was duplicated within the Clinical Laboratories. Information Systems The laboratory information system (LIS) used by the SFGH Clinical Laboratories was Sunquest, version 4.3 (Sunquest Information Systems, Tucson, AZ). A patient care information system (PCIS; CareVue, Hewlett-Packard, Palo Alto, CA) was used in the ICU. The LIS and PCIS were directly interfaced so that laboratory data could be viewed instantaneously after remote analysis was complete. Working Group The decision-making group ( working group ) charged with developing and implementing the plan for the mandated $35, budgetary savings for the clinical laboratories included representatives from Anesthesia and Peri- Operative Services, Respiratory Care Services, and Clinical Laboratories. Assessment and Planning The working group first determined that only blood gases (ph, PCO 2, PO 2 ) and ionized calcium values were of sufficient medical importance to warrant a 1-minute or less turnaround time and, thus, were worthy of consideration for POC testing. For the other tests that had been offered by the ICU laboratory (ie, Na+, K+, lactate, cooximetry), the group agreed that a 3-minute turnaround time was clinically acceptable; this decision permitted consolidation of these tests within the main Clinical Laboratories. Technical evaluation thus focused only on the performance of each commercially available POC test device for determination of ph, PCO 2, PO 2, and ionized calcium values, despite the broader testing capability of many of the POC devices. Phase 2: Technical Evaluation and Instrument Selection Instrumentation and Method Evaluation POC test systems evaluated included the i-stat (i- STAT, Princeton, NJ), IRMA (Diametrics Medical, St Paul, MN), and Opti-1 (AVL Scientific, Roswell, GA). All specimens were obtained by respiratory therapists using SIMS dry line draw ABG sampling kits (part 443W-2, SIMS Portex, Keene NH). All values obtained from POC devices were compared with those obtained with the same specimens analyzed in parallel with the relevant in-house method (for blood gases, AVL 995 or AVL Omni, AVL Scientific; for ionized calcium, Ciba-Corning 288 Blood Gas Analyzer, Ciba-Corning Diagnostics, Medfield, MA). Each of the POC devices used single-use disposable cartridges. Two i-stat cartridges were evaluated: G3+ (blood gas analysis only) and the EG7+ (blood gas analysis plus electrolytes, including ionized calcium). In 1995, ionized calcium test capability was available on the i-stat and IRMA, but only the i-stat method had been American Society of Clinical Pathologists Am J Clin Pathol 2;114: on 3 January 218

3 Ng et al / POINT-OF-CARE BLOOD GAS TESTING approved by the US Food and Drug Administration (FDA) for clinical use. During the technical evaluation phase, all POC test devices were located within the Clinical Laboratories immediately adjacent to the instruments used for the inhouse method. Commercially available control materials at 2 levels were used to determine intra-assay (n = 8-1) and interassay (n = 5-7) precision. From 42 to 57 patient specimens were analyzed in parallel with the in-house method and each of the 3 POC test systems; however, ionized calcium levels were determined in parallel with only 12 patient specimens on the IRMA, i-stat, and Ciba- Corning 288. Linear regression analysis was performed for each set of paired values (Excel, Microsoft, Redmond, WA). Data Analysis and Instrument Selection Values obtained by each of the POC devices were comparable to values obtained from the same specimen but analyzed with the in-house method Figure 1 Table 1. There were no obvious differences in technical performance between the individual POC devices. Selection of a POC device for SFGH thus was independent of technical performance and instead dependent on criteria unique to SFGH. Two of the POC devices (IRMA and Opti-1) were eliminated immediately from consideration because they lacked an FDA-approved method for ionized calcium determinations. Thus, i-stat was chosen because it had an FDA-approved method for ionized calcium determinations. An additional attractive feature of the i-stat was that it was the only POC test system evaluated that had an existing LIS interface. Financial Analysis The financial analysis for i-stat implementation was favorable, with projected savings of approximately $235, the first year of POC test implementation and sustained savings of approximately $222,5 annually thereafter Table 2. Implicit in this proposal was permanent savings attributable to elimination of 4 of the 5 technical staff positions assigned to the ICU laboratory. Elimination of these positions was possible because testing would be performed by existing licensed respiratory therapists (whose licensure status satisfied requirements of the State of California for testing personnel for this group of tests). Since respiratory therapists historically had been responsible for obtaining specimens for analysis, delegating testing responsibility to them permitted maximal efficiency and the shortest possible test turnaround time. The working group also recognized that the POC test program would be cost-effective only if the volume of testing could be controlled, since the per test reagent cost for POC testing was considerably higher than that incurred by the main Clinical Laboratories. Thus, the financial plan for POC testing budgeted for only that subset of tests that truly required turnaround times of 1 minutes or less (5% or less of the test volume performed at that time by the ICU laboratory), of which only 5% or less would require a simultaneous ionized calcium level. Implicit in the plan was that the other specimens currently tested in the ICU laboratory but not requiring a turnaround time of 1 minutes or less (5% or more of the total specimens) would be tested in the main Clinical Laboratories using existing Clinical Laboratories staff. The remaining technical position in the ICU laboratory that was not eliminated was transferred to the Clinical Laboratories to assist with the projected increased workload. Phase 3: Implementation The financial plan was presented to and accepted by hospital administration. Respiratory Care Services and Clinical Laboratories LIS staff then spent approximately 1 person-hours testing and resolving interface problems related to use of a local area network to capture i-stat test results for display on the ICU PCIS, monitor POC test utilization, and capture quality control and billing data. The ICU laboratory technical staff members were transferred to fill vacant positions within the Clinical Laboratories, and their respective positions in the ICU laboratory were eliminated. Respiratory therapists were trained in use of the POC device, sample handling, and analysis by i-stat representatives, in quality control by Clinical Laboratories staff, and in trouble-shooting LIS-PCIS interface problems by Respiratory Care Services staff. Respiratory therapists were educated to ignore any values other than ph, PO 2, PCO 2, and ionized calcium when using the EG7+ cartridge (that also yielded Na+, K+, and hemoglobin values), and the LIS-PCIS interfaces were configured so that only ph, PO 2, PCO 2, and ionized calcium values were captured by the LIS and transmitted for display on the PCIS whenever an EG7+ cartridge was used. After training, the Clinical Laboratories established a continuous quality control and assurance program that relied on test values obtained by respiratory therapists. The Clinical Laboratories and Respiratory Care Services also jointly established an ongoing proficiency testing program for all testing personnel to satisfy regulatory requirements for POC testing. Budgetary control for consumable supplies was transferred formally to Respiratory Care Services in 1996 when POC testing was implemented fully. 13 Am J Clin Pathol 2;114: American Society of Clinical Pathologists on 3 January 218

4 Clinical Chemistry / ORIGINAL ARTICLE A ph (i-stat) Delta ph (AVL-i-STAT) ph (IRMA) Delta ph (AVL-IRMA) ph (Opti-1) Delta ph (AVL-Opti) American Society of Clinical Pathologists Am J Clin Pathol 2;114: on 3 January 218

5 Ng et al / POINT-OF-CARE BLOOD GAS TESTING B Po 2 (mm Hg; i-stat) Delta Po 2 (mm Hg; AVL-i-STAT) Po 2 (mm Hg; AVL 995) Po 2 (mm Hg; AVL 995) Po 2 (mm Hg; IRMA) Delta Po 2 (mm Hg; AVL-IRMA) Po 2 (mm Hg; AVL 995) Po 2 (mm Hg; AVL 995) Po 2 (mm Hg; Opti-1) Delta Po 2 (mm Hg; AVL-Opti) Po 2 (mm Hg; AVL 995) Po 2 (mm Hg; AVL 995) 132 Am J Clin Pathol 2;114: American Society of Clinical Pathologists on 3 January 218

6 Clinical Chemistry / ORIGINAL ARTICLE C Pco 2 (mm Hg; i-stat) Delta Pco 2 (mm Hg; AVL-i-STAT) Pco 2 (mm Hg; IRMA) Delta Pco 2 (mm Hg; AVL-IRMA) Pco 2 (mm Hg; Opti-1) Delta PCO 2 (mm Hg; AVL-Opti) American Society of Clinical Pathologists Am J Clin Pathol 2;114: on 3 January 218

7 Ng et al / POINT-OF-CARE BLOOD GAS TESTING D Ionized Calcium (mmol/l; i-stat) Delta Ionized Calcium (mmol/l; C288-i-STAT) Ionized Calcium (mmol/l; C288) Ionized Calcium (mmol/l; C288) Ionized Calcium (mmol/l; IRMA) Delta Ionized Calcium (mmol/l; AVL-IRMA) Ionized Calcium (mmol/l; C288) Ionized Calcium (mmol/l; C288) Figure 1 Comparative analysis of values obtained with the i-stat (i-stat, Princeton, NJ), IRMA (Diametrics Medical, St Paul, MN), and Opti-1 (AVL Scientific, Roswell, GA) point-of-care testing devices. A, ph values. B, PO 2 values. C, PCO 2 values. D, ionized calcium values. Dotted lines in the left panels represents the ideal correlation that would have been achieved had both methods under comparison yielded identical results ( line of identity ); solid lines in all panels represent the trendline obtained with linear regression analysis; delta refers to the difference in values obtained between the in-house reference method (AVL 995 [AVL Scientific] or C288 [Ciba-Corning 288 Blood Gas Analyzer, Ciba-Corning Diagnostics, Medfield, MA]) and the method under evaluation (i-stat, IRMA, or Opti-1). Phase 4: Monitoring Utilization and Quality Utilization POC test utilization was monitored continuously by Respiratory Care Services Figure 2. Total test utilization and POC test utilization for blood gas analysis or co-oximetry did not increase appreciably during the 3 years since implementation (Figure 2A, Figure 2B). An unexpected cost-savings benefit was the tremendous reduction in the number of ionized calcium tests performed (Figure 2C). Historically, and owing to existing instrumentation, ionized calciums levels had been obtained with every blood gas analysis in the ICU laboratory, regardless of whether the information was needed for patient care. Use of 2 i-stat cartridges, 1 without (G3+) and 1 with ionized calcium determination capability (EG7+), for POC testing restricted ionized calcium determinations to situations 134 Am J Clin Pathol 2;114: American Society of Clinical Pathologists on 3 January 218

8 Clinical Chemistry / ORIGINAL ARTICLE Table 1 Linear Regression Parameters for Blood Gas Values Obtained From Point-of-Care Testing Devices * Slope Intercept r s Y X ph i-stat vs AVL 995 (n = 51) IRMA vs AVL 995 (n = 43) Opti-1 vs AVL 995 (n = 57) AVL Omni vs AVL 995 (n = 16) Po 2 i-stat vs AVL 995 (n = 5) IRMA vs AVL 995 (n = 42) Opti-1 vs AVL 995 (n = 53) AVL Omni vs AVL 995 (n = 162) Pco 2 i-stat vs AVL 995 (n = 51) IRMA vs AVL 995 (n = 43) Opti-1 vs AVL 995 (n = 57) AVL Omni vs AVL 995 (n = 162) i-stat vs AVL Omni (8/98) (n = 43) Ionized calcium i-stat vs C288 (n = 12) IRMA vs C288 (n = 12) s Y X, standard error of the estimate. * Linear regression analyses were performed for each set of paired values. Values obtained from the AVL 995 (ph, PCO 2, PO 2 ) or C288 (ionized calcium) were independent variables against which values obtained with all other instruments were regressed. The AVL 995 was our in-house reference standard for blood gas analysis until approximately 1996, after which the AVL Omni became our in-house reference standard. Regression analysis for performance of AVL Omni vs AVL 995 was performed using paired data from patient specimens analyzed on both instruments during n is the number of specimens. i-stat, i-stat, Princeton, NJ; IRMA, Diametrics Medical, St Paul, MN; Opti-1, AVL 995, and AVL Omni, AVL Scientific, Roswell, GA; C288, Ciba-Corning 288 Blood Gas Analyzer, Ciba-Corning Diagnostics, Medfield, MA. Regression parameters for i-stat values when PCO 2 discrepancies were discovered in August 1998 (Figure 3; see text for discussion). Table 2 Overall and Sustained Savings From Implementation of Point-of-Care (POC) Blood Gas Testing and Closure of the Intensive Care Unit Laboratory * Year 1 Year 2 Year 3+ Expenses Fixed Purchase of POC devices, eg, infrared links, printers (22,77) Interconnectivity hardware (2,7) Service contracts (POC devices and PC) (1,5) (1,5) (1,5) Variable Disposable POC test cartridges (3,5) (3,5) (3,5) Liquid commercial quality control materials (5) (5) (5) Subtotal (57,97) (32,5) (32,5) Savings Fixed Nonreplacement of the C288 38, Personnel (4 FTEs) 225, 225, 225, Liquid commercial quality control materials for the C288 5, 5, 5, Reagent supplies for the C288 25, 25, 25, Subtotal 293, 255, 255, Overall savings Expenses (57,97) (32,5) (32,5) Savings 293, 255, 255, Net annual savings 235,3 222,5 222,5 FTE, full-time equivalent clinical laboratory technologist; PC, personal computer. * Personnel savings were calculated by using the midpoint of the salary range for licensed clinical laboratory scientists in our geographic area in 1995 and included benefits (additional 2% of salary). Values are given in US dollars. Values in parentheses are expenses. C288 is the Ciba-Corning 288 Blood Gas Analyzer, Ciba-Corning Diagnostics, Medfield, MA American Society of Clinical Pathologists Am J Clin Pathol 2;114: on 3 January 218

9 Ng et al / POINT-OF-CARE BLOOD GAS TESTING A B C No. of Ionized Calcium Tests No. of Co-oximetry Tests No. of Blood Gas Tests 3, 2,5 2, 1,5 1, , 1,5 1, 5 January March May July September November January March May ICU ClinLabs RCS OR Total July September November January March May July September November Figure 2 Utilization of blood gas (A), co-oximetry (B), and ionized calcium (C) tests. Point-of-care (POC) blood gas testing was implemented in April 1996, coincident with closure of the ICU laboratory. Test volume according to location of laboratory or POC service (ie, RCS and OR) is shown. ClinLabs, Clinical Laboratories; ICU, ICU laboratory; OR, operating room; RCS, respiratory care services. in which the information was truly necessary for patient care (as determined by the ordering physician). The dramatic reduction in ionized calcium testing allowed Respiratory Care Services to purchase the less expensive G3+ cartridges for the majority of testing performed. Of all POC tests performed, approximately 1% to 3% of cartridges failed to yield results owing to internal cartridge failure, and 3% to 13% of cartridges were rejected owing to user error, bad batteries, bad contacts, or inadequate storage. Many of the POC devices required replacement during their first year of use. No POC testing related adverse effects on patient morbidity or mortality rates were reported. Value of Continuous Quality Control Monitoring At the time of POC implementation, the manufacturer recommended that use of 2 levels of electronic controls every 8 hours of use would satisfy CLIA 88 requirements for quality control. The Clinical Laboratories, however, insisted on a more rigorous quality control and assurance program consisting of the following: (1) electronic controls (2 levels) every 8 hours of use, (2) 1 level of liquid control material each week (3 levels of liquid controls total), and (3) weekly analysis of 1 patient specimen by the i-stat and the in-house reference method (ie, patient comparison specimen). Respiratory Care Services agreed to perform this more rigorous quality control and assurance program. Phase 5: Discontinuation of the i-stat In August 1998, 2 patient comparison specimens yielded discrepant PCO 2 values, with the i-stat value 15 to 2 mm Hg lower than that obtained with the reference in-house method. These discrepancies were of such magnitude that artificial ventilation parameters would have been changed. During this same period, liquid and electronic controls yielded values within acceptable ranges. Additional patient comparison specimens then were performed on 2 POC devices, of which 3 specimens yielded significantly different PCO 2 values. The PCO 2 discrepancies still were noted after testing with a new lot of disposable cartridges and software upgrades in the i- STAT devices Figure 3 (Table 1). The i-stat PCO 2 values, when discrepant, usually were lower than values obtained with the reference in-house method. The PCO 2 discrepancies were observed sporadically and were not attributable to any single POC device or any underlying diseases or therapies for the patients whose blood specimens yielded the discrepant PCO 2 values. In September 1998, following our inability to identify the cause of the PCO 2 discrepancies and growing lack of confidence in PCO 2 values obtained with the i-stat, all i-stat devices were removed from clinical use. The PCO 2 discrepancies were reported immediately to the FDA (MEDWATCH Program) and investigated on-site by i- STAT. In February 1999, the PCO 2 discrepancies were linked to an incompatibility between the sampling syringes used in our ICUs (SIMS dry line draw ABG sampling kit) and the i-stat PaCO 2 method (i-stat technical bulletin ). This incompatibility demonstrated by the manufacturer, however, did not reflect our actual experience. The manufacturer not only demonstrated discrepant PCO 2 values on a consistent basis, but also used experimental methods that did not reflect the actual clinical practice for obtaining and testing specimens. The cause of our observation of PCO 2 discrepancies on a seemingly random basis was never fully explained to our satisfaction. 136 Am J Clin Pathol 2;114: American Society of Clinical Pathologists on 3 January 218

10 Clinical Chemistry / ORIGINAL ARTICLE A 9 B Pco 2 (mm Hg; i-stat) Delta Pco 2 (mm Hg; AVL-i-STAT) Pco 2 (mm Hg; AVL Omni) Pco 2 (mm Hg; AVL Omni) Figure 3 Discrepant PCO 2 values obtained with the i-stat (i-stat, Princeton, NJ) were discovered serendipitously in August 1998 when 2 patient specimens yielded clinically significant discrepant values. Discrepant PCO 2 i-stat values were still observed after testing additional patient specimens (diamond, n = 11) using a new lot of test cartridges (triangle, n = 17) and following an upgrade of the i-stat software (, n = 15). Linear regression analysis was performed using values obtained with the AVL Omni (AVL Scientific, Roswell, GA) as the independent variable. A, Comparative analysis of PCO 2 values obtained with the i-stat and Omni (AVL Scientific). Dotted line represents line of identity. B, Delta (or difference) in PCO 2 values obtained between the in-house reference method (AVL Omni) and the i-stat. Discussion This report summarizes the many facets of our experience with implementation of i-stat POC blood gas testing in an acute care hospital. The impetus to develop this service was linked directly to budget cuts for the Clinical Laboratories with the mandate that the same level of service be maintained. The technical performance of all 3 POC devices evaluated (i-stat, IRMA, Opti-1) was reasonably comparable to our in-house methods, consistent with previous reports. 2-5 Implementation of the POC testing program allowed the Clinical Laboratories to achieve the mandated budget cut for The sustained savings achieved thereafter with this POC program, however, would not have been possible without the ongoing interdisciplinary collaboration of Anesthesia and Peri-Operative Services, Clinical Laboratories, and Respiratory Care Services at SFGH. Previous analyses of relative costs for POC testing vs costs for testing performed in centralized Clinical Laboratories demonstrated significantly higher reagent costs associated with POC testing. 6 Although there was concern that the ease and ready access of POC testing invariably would result in higher test utilization and ultimately higher variable costs, 7 we have not observed an increase in overall utilization or variable costs. Assigning complete fiscal responsibility to those who were responsible for ordering the tests and using the results for patient management (ie, Anesthesia and Peri-Operative Services and Respiratory Care Services) was the stimulus for regular monitoring of test utilization and for curbing periodic excessive and inappropriate use. Similar and perhaps greater savings might have been achieved if rapid specimen transport (eg, dedicated messenger service or pneumatic tube with tracking and soft-landing capability) would have been available so that all testing could have been consolidated within the main Clinical Laboratories. 8 The ongoing collaborative effort among the 3 affected SFGH services also was critical for ensuring continued compliance with various regulatory requirements for POC testing personnel performed under the licensure of a clinical laboratory. 9 An unexpected finding of this collaborative venture was the priceless value of an ongoing collaborative quality control and assurance program that exceeded manufacturer s recommendations. The early experience with these POC devices suggested that they were highly accurate and precise, American Society of Clinical Pathologists Am J Clin Pathol 2;114: on 3 January 218

11 Ng et al / POINT-OF-CARE BLOOD GAS TESTING warranting a lower level of quality oversight. In 1996, i-stat in fact received confirmation from the Health Care Financing Administration that testing with an electronic simulator every 8 hours fully met the CLIA 88 requirements for quality control. The PCO 2 discrepancies we discovered, however, would not have been detected had we relied solely on this minimal level of quality control. Furthermore, our PCO 2 discrepancy problem would not have been detected with our ongoing testing of commercial liquid controls a quality control measure exceeding manufacturer s recommendations. It was only with the continuous comparative analysis of patient specimens (yet another layer of quality control exceeding manufacturer s recommendations) that we discovered the clinically relevant and serious problem with PCO 2 discrepancies. The rapid detection of the PCO 2 discrepancies was because of the ever vigilant continuous monitoring of patient comparison specimen results by Respiratory Care Services personnel. The close collaboration among the 3 services allowed rapid investigation of possible causes for the observed PCO 2 discrepancies and the decision to stop all POC blood gas testing when the cause could not be identified quickly, thereby averting potentially adverse medical outcomes. Although the PCO 2 discrepancies we observed were attributed to an incompatibility between the sampling syringes and the i-stat PaCO 2 method, there is still no explanation for our observation of these discrepancies on a sporadic basis. Systematic biases for blood gas values have been observed previously with different commercial sources of sampling syringes, 1 but sporadic discrepancies such as we observed have not been reported as solely attributable to sampling syringes. Addendum to the i-stat Experience The investigation into the i-stat PCO 2 discrepancies was not completed formally until the spring of 1999, at least 5 months after we discontinued use of the i-stat devices. Although the working group agreed to cease all use of i- STAT devices, Anesthesia and Peri-Operative Services and Respiratory Care Services still required rapid blood gas analysis. Of the 2 remaining POC devices available, the working group agreed to implement the newer version of Opti-1 (Opti-CCA). Formal test method validation and verification was completed in April 1999, purchase and contract negotiations in June 1999, and full implementation in July During the interim when no POC blood gas analysis was available, all testing was performed within the main Clinical Laboratories with Respiratory Therapists providing rapid transport of samples from the ICUs. Anesthesia, University of California San Francisco and Respiratory Care Services, San Francisco General Hospital, San Francisco, CA. Presented in part as a lecture and poster at the American Association for Respiratory Care, 45th International Congress, Las Vegas, NV, December 14, Address reprint requests to Dr Ng: Clinical Laboratories, NH-2M2, San Francisco General Hospital, 11 Potrero Ave, San Francisco, CA * Ms Hogan is currently with Carol M. Hogan Associates, Palo Alto, CA. Acknowledgments: We are grateful for the expert advice and consultative comments of Richard Schlobohm, MD, and Hassan Khayam-Bashi, PhD, and the expert technical assistance of Maggie Ziman, MT, Margaret Fong, MT, and staff of the Hematology Division of the San Francisco General Hospital Clinical Laboratories and of the San Francisco General Hospital Respiratory Care Services. References 1. De Cresce RP, Lifshitz MS, Logue LJ. Managed care and the hospital laboratory: survival of the fittest. Clin Lab Manage Rev. 1994;8: Murthy JN, Hicks JM, Soldin SJ. Evaluation of i-stat portable clinical analyzer in a neonatal and pediatric intensive care unit. Clin Biochem. 1997;3: Schneider J, Dudziak R, Westphal K, et al. The i-stat analyzer: a new, hand-held device for the bedside determination of hematocrit, blood gases, and electrolytes [in German]. Anaesthesist. 1997;46: Prause G, Kaltenbock F, Doppler R. Preclinical blood gas analysis, 2: experience with three blood gas analyzers in emergency care [in German]. Anaesthesist. 1998;47: Zaloga GP, Roberts PR, Black K, et al. Hand-held blood gas analyzer is accurate in the critical care setting. Crit Care Med. 1996;24: Nosanchuk JS, Keefner R. Cost analysis of point-of-care laboratory testing in a community hospital. Am J Clin Pathol. 1995;13: Bachner P. Alternate site testing: the old and new paradigm or the past is prologue. Arch Pathol Lab Med. 1995;119: Keshgegian AA, Bull GE. Evaluation of a soft-handling computerized pneumatic tube specimen delivery system: effects on analytical results and turnaround time. Am J Clin Pathol. 1992;97: Belanger AC. Alternate site testing: the regulatory perspective. Arch Pathol Lab Med. 1995;119: Carter BG, Tibballs J, Hochmann M, et al. A comparison of syringes to collect blood for analysis of gases, electrolytes and glucose. Anaesth Intensive Care. 1994;22: From the 1 Department of Laboratory Medicine, University of California San Francisco; 2 Clinical Laboratories, San Francisco General Hospital, San Francisco, CA; and 3 Department of 138 Am J Clin Pathol 2;114: American Society of Clinical Pathologists on 3 January 218