Implementation of Point-of-Care Rapid Urine Testing for Drugs of Abuse in the Emergency Department of an Academic Medical Center

Transcription

1 Clinical Chemistry / POC Testing f o r Dr u g s o f Ab u s e in t h e ED Implementation of Point-of-Care Rapid Urine Testing for Drugs of Abuse in the Emergency Department of an Academic Medical Center Impact on Test Utilization and ED Length of Stay Kent Lewandrowski, MD, 1 James Flood, PhD, 1 Christine Finn, MD, 2 Bakhos Tannous, PhD, 1 Alton B. Farris, MD, 1 Theodore I. Benzer, MD, PhD, 3 and Elizabeth Lee-Lewandrowski, PhD, MPH 1 Key Words: Drugs of abuse; Point of care; Toxicology; Emergency department; Length of stay; Turnaround time DOI: /59681R72JDTCCD2B Abstract We evaluated the impact of implementing a pointof-care (POC) rapid urine test panel for drugs of abuse on turnaround time (TAT), emergency department length of stay (LOS), and laboratory test utilization patterns. The mean TAT from sample collection to results reporting decreased by 69.4%, from 108 to 33 minutes, the mean LOS from 11.1 to 8.1 hours (27%; P <.0001), and the median LOS from 8.0 to 7.0 hours (13%; P =.0017). A method crossover between the POC and central laboratory methods revealed differences in sensitivity and specificity. Overall, there was no clear preference for either method. Differences in performance for individual drug classes were reconciled by providing interpretive comments with POC results. Following implementation, use of urine testing for drugs of abuse increased by 30%, which was offset by fewer requests for extended toxicology testing in the central laboratory and more selective ordering of toxicology tests not on the POC panel (alcohols and analgesics). The implementation of a POC urine test panel for drugs of abuse decreased LOS and TAT and essentially replaced central laboratory testing for drugs of abuse. Differences in sensitivity and specificity between POC and central laboratory results require provision of interpretive comments with results. Emergency department (ED) overcrowding and prolonged patient length of stay (LOS) are increasing problems in most American hospitals. 1 In a recent study, ED overcrowding was reported in 62% of US hospitals and has been reported in other countries, including those in the United Kingdom, Australia, Canada, and Spain. 1 EDs in urban areas must adapt to overcrowding, high acuity, and the need to minimize LOS. 2,3 These problems have been highlighted at the national level, including reports in the lay press. 2 A variety of reasons for ED overcrowding have been cited, including episodic influenza epidemics, staffing shortages, high patient acuity, difficulties in obtaining timely specialist consultations, insufficient space, lack of available inpatient beds, delays in laboratory and radiologic tests, and other factors. 1-3 Although the operations of emergency services are complex, the availability of ancillary services, including the turnaround time for laboratory tests, may influence the time required to triage, diagnose, and make decisions about patient disposition. However, there are relatively few data on the impact of rapid point-of-care testing (POCT) on ED operations, including patient LOS. 3,4 Some types of rapid testing such as blood gases and electrolytes seem to have little impact on ED operations. 5,6 On the other hand, there is increasing evidence that POCT for cardiac markers (including troponin, creatine kinase [CK]-MB, myoglobin, and B-type natriuretic peptides) can reduce ED LOS and unnecessary hospital admissions. 3,7-10 The menu of laboratory tests now available in POCT formats is extensive and continually expanding. 11 This expansion of testing options potentially creates new opportunities to improve ED operations. The unit cost of POCT is generally higher than testing in a centralized laboratory. Managing a POCT program is also 796 Am J Clin Pathol 2008;129: DOI: /59681R72JDTCCD2B

2 Clinical Chemistry / Original Article challenging owing to difficulties of maintaining regulatory compliance. It is, therefore, important to identify which tests have a positive impact on patient outcomes or on ED operations to avoid adding cost without a demonstrable benefit. We report an evaluation of the impact of a rapid POCT urine test panel for drugs of abuse on patient LOS and test utilization in the ED of a large, urban, academic medical center. Materials and Methods The Massachusetts General Hospital (MGH), Boston, is an 899-bed university-affiliated medical center serving greater Boston and surrounding areas. The hospital admitted 49,404 inpatients and received more than 1 million visits to outpatient facilities in the 2006 fiscal year. The full service ED includes a level I trauma center. There were more than 79,000 ED visits in the 2006 fiscal year, averaging about 215 patients per day. The ED serves as the portal of admission for more than 40% of MGH inpatients. The ED has 3 triage stations, an 11-bed trauma/acute unit, a 17-bay major multipurpose unit for acute adult illness, a 4-bed pediatric unit, a 7-bed minor multipurpose unit for minor surgical and orthopedic procedures, a 7-bed rapid diagnostic unit, and an acute psychiatry service featuring a locked unit composed of 3 holding rooms, 4 interview rooms, and a secured waiting area. In the 2006 fiscal year, the average LOS was 6.7 hours (10.0 hours for patients admitted to MGH and 6.1 hours for patients admitted to the ED observation unit). Approximately 27% of patients seen in the ED were admitted to MGH, with an average length of time between bed request and assignment of 4 hours. The clinical laboratories operate a 24-hour ED satellite laboratory staffed by medical technologists located in the ED. The ED satellite laboratory POCT menu includes assays for CK-MB, troponin I, urinalysis, urine pregnancy testing (human chorionic gonadotropin), rapid group A streptococcus, rapid influenza A/B, and rapid respiratory syncytial virus. The main clinical laboratory operates a full-service, 24-hour toxicology menu, including serum and urine screens for drugs of abuse and extended toxicologic testing using gas and liquid chromatographic techniques. The extended toxicology test menu includes 5 panels (alcohols, panel 1; analgesics, panel 2; barbiturates and antiepileptic drugs, panel 3; benzodiazepines, panel 4; and tricyclics, cocaine, and other, panel 5). All 5 panels together comprise the extended (full toxicology) menu. The Biosite Triage (Biosite Diagnostics, San Diego, CA) rapid drugs of abuse testing system is an instrumentread, single-use, disposable cartridge multidrug fluorescence immunoassay panel. This approach avoids the need to visually read the test results on the cartridge and permits results to be interfaced directly to the laboratory information system. The urine test panel for drugs of abuse includes amphetamines, methamphetamines, barbiturates, benzodiazepines, cocaine metabolites, opiates, phencyclidine, and tetrahydrocannabinol (THC; cannabinoids). After completion of this study, urine testing for acetaminophen and methadone were added to the panel. Before implementation, the Triage system was compared with the established Syva EMIT II and II+ (Dade Behring, Newark, DE) urine test panel for drugs of abuse performed on the Hitachi 911 analyzer (Roche/Hitachi Diagnostics, Indianapolis, IN) in the central chemistry laboratory. Discrepant results between the 2 systems were reconciled using simultaneously collected serum samples analyzed using high-performance liquid chromatography 12,13 as the referee method. We measured 91 urine samples on the Biosite Triage and Hitachi 911 and compared the results (positive or negative). We tested each sample for the following 7 classes of drugs of abuse: amphetamines, barbiturates, benzodiazepines, cocaine metabolites, opiates, phencyclidine, and THC. Analysis of test turnaround time, test volumes, and ED LOS were performed by audit of the laboratory and hospital patient care information systems. Overall records were reviewed for 197 patients before POCT implementation and 286 patients after implementation. This study was approved by the institutional review board in our hospital. Results ztable 1z shows the results of the validation study of the Triage drugs of abuse panel compared with the established EMIT method in the clinical laboratory. Validation of the Triage tricyclic antidepressant test was reported in a separate study. 14 Owing to differences in analytic sensitivity (cutoff value) and interference from other drugs and metabolites, we expected some discrepant result between the methods. The important issue in this validation was to identify sources of discrepant results such that we could report the POCT results in a manner that could coexist with the reporting of results from the central laboratory. Overall, there were 39 discrepant results in 91 drugs of abuse panels (7 tests per panel, or 637 individual drug assays) yielding 6.1% discrepant results. The probable explanation for these discrepancies based on serum toxicology and selected chart reviews are listed in Table 1. We noted important differences between individual Triage and EMIT II tests. The Triage was more specific for amphetamines, methamphetamines, and phencyclidine. The EMIT II was more sensitive for cocaine metabolites, THC, and opiates. ztable 2z shows the interpretive caveats that we report with the POCT drugs of abuse results to harmonize interpretation of the POCT test with results originating from the central laboratory. We currently use the EMIT II+ amphetamines Am J Clin Pathol 2008;129: DOI: /59681R72JDTCCD2B 797

3 Lewandrowski et al / POC Testing f o r Dr u g s o f Ab u s e in t h e ED ztable 1z Comparison of the Biosite Triage Panel for Urine Testing for Drugs of Abuse With the Syva EMIT II Testing System in the Central Laboratory No./Total (%) EMIT II+/ Triage+/ Drug or Metabolite Discrepant Cases Triage EMIT II Comments Amphetamines 7/92 (8) 7 0 EMIT II false-positives due to use of chlorpromazine (3), promethazine Methamphetamines 1/92 (1) 0 1 (2), bupropion (1), and selegiline (1) One EMIT II false-negative due to use of Ecstasy (MDMA detected in serum) Barbiturates 0/92 (0) 0 0 Benzodiazepines 15/92 (16) EMIT II false-negatives due to use of lorazepam (11) and midazolam (1); 1 Triage false-negative due to use of clonazepam; 2 discrepancies Cocaine 3/93 (3) 3 0 unresolved Likely 3 Triage false-negatives: EMIT II more sensitive with 150-ng/mL cutoff and sensitive to 2 cocaine metabolites; Triage has 300-ng/mL cutoff and detects only 1 metabolite Opiates 5/93 (5) 5 0 Probable Triage false-negatives due to oxycodone use (all 5 cases) Phencyclidine 2/92 (2) 2 0 Both EMIT II false-positives due to dextromethorphan use THC (cannabinoids) 6/92 (7) 4 2 The 4 EMIT II+/Triage likely Triage false-negatives: EMIT II more sensitive with a 20-ng/mL cutoff and more sensitive to several THC metabolites; Triage has 50-ng/mL cutoff; The 2 Triage+/EMIT II likely Triage false-positives due to pantoprazole use MDMA, 3,4-methylenedioxymethamphetamine; THC, tetrahydrocannabinol. ztable 2z Interpretive Comments Appended to Reports for Positive and Negative Results of Urine Testing for Drugs of Abuse by the Triage Testing System Result Positive results Benzodiazepines Opiates Tetrahydrocannabinol Tricyclics Negative results Tricyclics Comment This ED laboratory assay detects lorazepam (Ativan) use. The central laboratory urine benzodiazepines test generally does not. This assay is not sensitive for detection of oxycodone and oxymorphone. False-positives may be caused by use of pantoprazole (Protonix). True-positive ED laboratory urine tricyclic antidepressant results are associated with subtherapeutic and higher serum concentrations of amitriptyline, nortriptyline, imipramine, desipramine, and doxepin. False-positive results can be caused by use of cyclobenzaprine (Flexeril). False-negative ED laboratory urine tricyclic antidepressant results are associated with clomipramine (Anafranil) use. ED, emergency department. assay, which has been shown to be more specific than its predecessor, EMIT II. 15 The mean turnaround time for urine testing for drugs of abuse from the time of specimen collection until results reporting was reduced from 108 minutes before POCT to 33 minutes after POCT (69.4% reduction). zfigure 1z shows the mean and median ED LOS for patients who received drugs of abuse testing before (n = 197 patients) and after (n = 286 patients) implementation of POCT of urine for drugs of abuse in our ED satellite laboratory. Overall, the mean ED LOS declined from 11.1 to 8.1 hours (P <.0001; unpaired t test) and the median ED LOS from 8.0 to 7.0 hours (P =.0017; Mann-Whitney U test). zfigure 2z shows the volume of requests for urine testing for drugs of abuse from the ED satellite laboratory and the central laboratory over time. Before implementation of POCT, we averaged 11.9 panels per day in the central laboratory from the ED. Following implementation, the number of urine panels for drugs of abuse performed for the ED in the central laboratory declined to fewer than 1 per day within approximately 4 months, and the number of tests performed by LOS (h) Before Mean P <.0001 After Mean Before Median P =.0017 After Median zfigure 1z Mean and median emergency department length of stay (LOS) before and after implementation of point-of-care urine testing for drugs of abuse. 798 Am J Clin Pathol 2008;129: DOI: /59681R72JDTCCD2B

4 Clinical Chemistry / Original Article Lab POCT No. of Test Panels zfigure 2z Number of urine testing panels performed for drugs of abuse in the central laboratory and by point-of-care testing (POCT) over time. POCT increased and essentially replaced central laboratory testing. By the end of the study period, there was a small increase in test utilization to 15.5 panels per day at 1 month following implementation. It is not clear if this increase in utilization was appropriate or reflected excess utilization. The increase in urine panels for drugs of abuse requested was offset by a reduction in extended serum toxicology panels performed in the central laboratory from 67.7% of all ED toxicology requests to 53.6%. In contrast, requests for selected panels 1 (alcohols) and 2 (analgesics) increased from 30.6% of all toxicology requests to 45.5%. Collectively, this pattern of utilization reflects increasing use of urine testing for drugs of abuse (augmented with serum panels 1 and 2) coupled with a decrease in requests for full extended serum toxicology testing. Discussion The literature documenting the impact of rapid POCT on ED operations and outcomes (such as LOS) has produced mixed results. For example, studies on POCT for blood gases and electrolytes in the ED showed no impact on LOS. 5,6 In recent years, rapid whole blood cardiac marker testing, including CK-MB, troponin, and B-type natriuretic peptides, has been shown to reduce ED LOS and unnecessary hospital admissions However, POCT is generally more expensive than central laboratory testing and may be challenging to manage, particularly in regard to maintaining regulatory compliance as mandated by the Clinical Laboratory Improvement Amendments of 1988 and the Joint Commission on Accreditation of Healthcare Organizations. For these reasons, it is important to define which POC tests contribute to improved outcomes and to avoid implementing POCT that only adds to cost and managerial overhead without a demonstrable benefit to patient care or ED operations. In this study, we evaluated the impact of implementing rapid POCT of urine for drugs of abuse in the ED of a large, urban, academic medical center. At MGH, we performed testing in our ED satellite laboratory that is staffed 24 hours a day by trained medical technologists. The satellite laboratory performs a menu of tests, including urine pregnancy testing, dipstick urinalysis, whole blood cardiac markers, influenza A/B, respiratory syncytial virus, and rapid streptococcus A testing. The rapid urine test panel for drugs of abuse was shown to reduce test turnaround time by 69.4% compared with central laboratory testing and was associated with a statistically significant decrease in the mean and median ED LOS for patients who required urine testing for drugs of abuse. The decrease in mean LOS of 3 hours multiplied by the number of tests per day (15.5 panels) translates into 46.5 bed hours, or approximately 1.94 bed days, saved per day. With the median LOS, the estimate would be 0.65 bed days saved per day. Based on our results, it is clear that implementation of urine testing for drugs of abuse produced a substantial benefit to ED operations. An important caveat concerns the value of measuring mean vs median LOS. The value for the mean LOS reduction was much greater than the median value owing to the impact of LOS outliers. For this reason, it is important to calculate both values to assess the impact on LOS. Implementation of POCT often produces changes in test utilization patterns (usually increased utilization). During a period of 2 months after implementation, the rapid POCT panel essentially replaced requests from the ED for urine testing for drugs of abuse in the central laboratory. We Am J Clin Pathol 2008;129: DOI: /59681R72JDTCCD2B 799

5 Lewandrowski et al / POC Testing f o r Dr u g s o f Ab u s e in t h e ED observed an increase in orders for urine testing for drugs of abuse from 11.9 to 15.5 panels per day (30% increase in utilization). This increase was offset by a decrease in the percentage of requests for the extended toxicology (5-panel) test and an increase in selective ordering of panels 1 (alcohols) and 2 (analgesics). After the study was completed, Biosite Diagnostics added methadone and acetaminophen to the test panel. We have not collected data on the impact of these additions on test utilization. Panels for urine testing for drugs of abuse from different manufacturers differ in cutoff values used to determine a positive test result and in the spectrum of other drugs and drug metabolites that may produce false-positive results. Our crossover evaluation comparing the Biosite Triage with the established Syva EMIT II method revealed a number of differences between the methods, with an overall 6.1% rate of discordant results. On balance, we did not conclude either method was superior to the other, although significant differences were observed for individual drug classes. These results demonstrate the importance of performing a careful crossover validation study and of establishing interpretive comments that are included with test results to harmonize drugs of abuse testing when using more than 1 method in 2 locations. Implementation of POCT usually involves incremental cost to the institution. 3 Most studies have reported that the POCT option is more expensive on a unit-cost basis than equivalent testing performed in the central laboratory. 3 However, comparison of unit costs fails to consider the impact of POCT on the overall systems cost of providing clinical care. In the case of POCT in the ED, other factors must be considered, such as ED LOS, changes in test utilization, and the impact of testing on ED workflow. The impact of POCT on revenues is even more complex. 3 If POCT reduces ED LOS, presumably more patients can be evaluated, and revenues would increase. The amount of revenue will depend on the payer mix and the amount of free care provided by the ED. Furthermore, some patients will be admitted to the hospital, and these revenues and costs need to be factored into the analysis. In short, performing simple cost analyses of POCT is highly artificial when testing is implemented in a complex operation in which costs and revenues need to be considered. The unit cost of a urine test for drugs of abuse in our ED laboratory, including consumables and labor, is $ The comparable cost in the central laboratory for urine testing for drugs of abuse is $9.40. Given a test volume of 15.5 POC urine tests for drugs of abuse per day, the total annual cost for the service is $132,951. Of this, 3.6 panels per day represented an increase in utilization. This increase was offset in part by a reduction in central laboratory costs for urine testing for drugs of abuse of $40,829 and by replacement of some extended toxicology tests by ordering of individual panels. The latter figure is difficult to quantify. Therefore, the incremental cost of the POCT option is $92,122. The median reduction in ED LOS was 1 hour per test. Assuming 15.5 tests per day, a total of 15.5 hours of ED bed time per day was saved, which is equivalent to 0.65 ED beds per 24 hours. The cost of an ED bed in our institution is approximately $55 per hour. Annualized, this equates to hypothetical savings of $313,170. Based on this simplistic analysis, it seems that the POCT option is cost-effective. However, the actual savings is impossible to calculate because the 0.65 beds saved were not actually eliminated, nor was the infrastructure needed to maintain it. The impact on ED revenues and the costs and revenues arising from inpatient admissions from the ED is equally elusive. Because of these complications, we cannot resolve the question of whether implementing POC urine testing for drugs of abuse is cost-effective. Rather, the question is whether the 0.65 beds that were made available for new patients is worth the $92,122 per year in incremental cost to the laboratory. Viewed from this perspective, the decision of whether to continue urine testing for drugs of abuse in our ED is largely intuitive. From the Departments of 1 Pathology and Laboratory Medicine, 2 Psychiatry, and 3 Emergency Services, Massachusetts General Hospital and Harvard Medical School, Boston. Funded in part by a quality assurance grant from Biosite Diagnostics. Address reprint requests to Dr Lee-Lewandrowski: Gray 5 Chemistry, Massachusetts General Hospital, Fruit Street, Boston, MA References 1. Bradley V. Placing emergency department crowding on the decision agenda. Nurs Econ. 2005;1: Landro L. Emergency rooms hope new technology will ease their trauma. Wall Street Journal. July 13, 2001:B1. 3. Lee-Lewandrowski E, Corboy D, Lewandrowski K, et al. Implementation of a point-of-care satellite laboratory in the emergency department of an academic medical center: impact on test turnaround time and patient emergency department length of stay. Arch Pathol Lab Med. 2003;127: Murray RP, Leroux M, Sabga E, et al. Effect of point of care testing on length of stay in an adult emergency department. J Emerg Med. 1999;17: Parvin C, Lo S, Deuser S, et al. Impact of point-of-care testing on patient s length of stay in a large emergency department. Clin Chem. 1996;42: Kendall J, Reeves B, Clancy M. Point-of-care testing: randomized controlled trial of clinical outcome. BMJ. 1998;316: Singer A, Ardise J, Gulla J, et al. Point-of-care testing reduces length of stay in emergency department chest pain patients. Ann Emerg Med. 2005;45: Blick K. Economics of point-of-care testing for cardiac markers and B-natriuretic peptide. Point of Care. 2005;4: Am J Clin Pathol 2008;129: DOI: /59681R72JDTCCD2B

6 Clinical Chemistry / Original Article 9. Mueller C, Scholar A, Laule-Kilian K, et al. Use of B-type natriuretic peptide in the evaluation and management of acute dyspnea. N Engl J Med. 2004;350: Green S, Redmond P, Januzzi J, et al. The impact of amino-terminal pro-brain natriuretic peptide testing on hospital length of stay and morbidity in patients with acute decompensated heart failure. Arch Pathol Lab Med. 2007;131: Lewandrowski K, Pagnani G, Walsh M, et al. Evolution of point-of-care testing in a large academic teaching hospital. Point of Care. 2005;4: Puopolo P, Pothier M, Volpicelli S, et al. Single procedure for detection, confirmation, and quantification of benzodiazepines in serum by liquid chromatography with photodiode-array detection. Clin Chem. 1991;37: Puopolo P, Volpicelli S, Johnson D, et al. Emergency toxicology testing (detection, confirmation, and quantification) of basic drugs in serum by liquid chromatography with photodiode array detection. Clin Chem. 1991;37: Melanson S, Lee-Lewandrowski E, Griggs D, et al. Interpreting tricyclic antidepressant measurements in urine in an emergency setting: comparison of two qualitative point-ofcare urine tricyclic antidepressant drug immunoassays with quantitative serum chromatographic analysis. J Anal Toxicol. 2007;31: Melanson SF, Lee-Lewandrowski E, Griggs DA, et al. Reduced interference by phenothiazines in amphetamine drug of abuse immunoassays. Arch Pathol Lab Med. 2006;130: Am J Clin Pathol 2008;129: DOI: /59681R72JDTCCD2B 801