Evaluation of Stat and Routine Turnaround Times as a Component of Laboratory Quality

Transcription

1 Evaluation of Stat and Routine Turnaround Times as a omponent of Laboratory Quality LEE H. HILBORNE, M.D., ROBERT K. OYE, M.D., JOSEPH E. McARDLE, M.D., JEFF A. REPINSKI, M.S., MT(ASP), AND DENIS O. RODGERSON, PH.D. Quality assurance has been an essential part of clinical laboratory operations for more than two decades. Analytic precision and accuracy goals have been established, and laboratory performance is monitored periodically. For a laboratory test to be useful, it must be available in a timely manner. Expedience of result reporting has not, however, been routinely included among measures of laboratory quality. The authors evaluated stat and routine turnaround times for 42,414 requests on 24 clinical analytes over a 14-day period with the use of a personal computer. Median turnaround time is 1.70 times faster for stat than for routine tests. When examined by shift, average turnaround time is considerably faster for tests ordered stat than for tests ordered routinely during the day and evening shifts, when the work load is the greatest. The authors are now examining turnaround time as an indicator of quality laboratory performance and efficiency. omputer systems in many clinical laboratories already have the sophistication necessary to perform turnaround time analysis. The authors recommend that clinical laboratories begin to include timeliness of stat and routine result reporting as part of their quality assurance programs. Laboratories may also wish to investigate the usefulness of stat requests during slower shifts because these requests may interrupt the normal flow of specimens without expediting result reporting. (Key words: Turnaround time; Stat tests; Routine tests; Quality assurance; omputerization; Efficiency) Am J lin Pathol 1989;91: ASSURANE OF QUALITY has been an expanding component of clinical laboratory operations since proficiency testing began in Participation in proficiency testing programs became mandatory with the passage of the linical Laboratory Improvement Act of " Major quality assurance activities for clinical laboratories include annual inspection to ensure compliance with established guidelines, participation in an w/erlaboratory proficiency testing program to evaluate accuracy of test performance compared with an established or consensus value, and /n'ralaboratory accuracy and precision studies for each analyte performed. The ollege of American Pathologists' (AP) proficiency testing program is the largest Received April 25, 1988; received revised manuscript and accepted for publication June 14, Supported in part by a grant from the Robert Wood Johnson Foundation. The opinions and conclusions herein are those of the authors and do not necessarily represent the views of the Robert Wood Johnson Foundation. Address reprint requests to Dr. Hilborne: Robert Wood Johnson linical Scholars Program, Department of Pathology, University of alifornia Medical enter, Los Angeles, alifornia Departments of Pathology and Medicine, University of alifornia, Los Angeles, Medical enter, Los Angeles, alifornia of its kind, and there were approximately 4,000 participants enrolled in the 1987 AP Laboratory Accreditation Program. 3 Despite widespread participation, however, and the thoroughness of testing procedures, timeliness of result reporting is not routinely included among standard measures of laboratory quality. omputer sophistication and electronic data-recording equipment have dramatically increased in the health care industry with extensive advances in computer hardware and software, particularly for clinical laboratory applications. Many laboratories can now electronically follow specimens from receipt in the laboratory to the time analysis is completed and released to the clinician. Some advanced hospital laboratory computer systems can track specimens from the time of order entry on the patient's hospital floor. Laboratory computerization now allows us to look closely at laboratory performance based on the time it takes to process, analyze, and report a test result. This is the first report, to our knowledge, that quantitates and evaluates stat and routine turnaround times. We present our laboratory turnaround times and how we have started to use these measures as an indicator of quality of laboratory performance and laboratory efficiency. Methods Our hospital uses a ommunity Health omputing (H) (Houston, TX) laboratory information system (LIS) with Stratus hardware. We transferred data to floppy diskettes in an MS-DOS format and then analyzed them with an IBM-AT personal computer using the SAS statistical package (SAS, Inc., ary, N). The search of our LIS initially captured the specimen type (e.g., sodium, potassium, activated partial thromboplastin time), work area (e.g., automated chemistry, hematology, emergency room laboratory, clinic laboratory), specimen request time, specimen "in-laboratory" time, specimen completion time, and specimen priority (i.e., stat or routine). The shift (i.e., day, evening, night) is classified by specimen 331

2 332 HILBORNE ET AL. A.J..P. March 1989 completion time. We define turnaround time as the difference between the specimen in-laboratory and completion times. Included were all tests that our laboratory generally offers on a stat or routine basis with the exceptions that follow. We excluded data when there was evidence that receipt or completion times may have been altered because of subsequent requests on the same specimen (i.e., the request time is later than the in-laboratory time). Specimens analyzed and completed during regularly scheduled computer maintenance periods were also excluded because recorded times may not accurately reflect specimen turnaround time. We also excluded those tests that often require verification by a supervisor or pathologist because, although these results are usually conveyed to the clinician in a timely manner, our quality control protocol requires supervisor verification, which alters the final specimen completion time; these specimen types include white blood cell differentials, body fluid cell counts, and Gram's stains. Arterial blood gases were not included in the study because the urgency of these requests often precludes specimen accessioning before completion and reporting of the results. Analytes ordered as part of an outpatient chemistry panel are batch processed and were therefore excluded from this study. Other than panel testing, it is not possible for the medical technologist to readily identify a specimen as coming from an inpatient or outpatient source; therefore, no attempt was made to separate inpatient from outpatient specimens. We also excluded from our analysis one day when our LIS was down for seven hours. For this analysis, tests were grouped together, based on similar methods, instrumentation, and work area. Table 1 indicates the members of each group and the total number of assays performed per analyte. The "basic-7" analytes were grouped accordingly because they are often ordered together and can be assayed by many instruments throughout our medical center. Enzymes studied included only those three that are offered on a routine and stat basis on all shifts without pathologist approval. The emergency room (ER) and perinatal (PNL) laboratories are tabulated together because they are expected to have faster turnaround times since they have smaller test volumes, perform fewer analytes, and are located near the facilities and physicians they primarily serve. Although courier transit time clearly makes a major contribution to turnaround time, the current sophistication of our LIS does not permit order entry from the patient's location; we excluded this component of turnaround time from our analysis. The clinic laboratory included in our study is in our medical oncology outpatient center. Among those analytes we studied, only hematology tests are performed at this location. Other satellite laboratories within the medical center do not routinely perform those tests included in our study. Turnaround time does not follow a normal (Gaussian) distribution, so parametric statistics cannot be used for analysis. Therefore, we chose to look at median turnaround times for comparison of routine and stat testing. "Stat percent faster" in Tables 2 and 3 is defined as follows: Median turnaround time (routine) - median turnaround time (stat) 77T T~- ; T Median turnaround time (stat) Results x 10 We analyzed turnaround times from 50,198 test results from which we obtained 42,414 usable values; this represents all requests for the specified analytes for 14 days, with the exceptions noted in the "Methods" section. Our personal computer with an 8-MHz processor took approximately 120 minutes to complete the analysis when provided with the raw data in an ASII text file. Once the data were provided, no additional user interaction was necessary before the data were printed in tabular form. We performed the same analysis with our hospital IBM Model 3090 mainframe computer, which completed the processing in less than 30 seconds. We studied turnaround times for tests by group, shift, and specimen priority (Table 2) and by work area and priority (Table 3). Data from Table 2 are shown graphically in Figure 1. Table 2 shows the median turnaround time in minutes by priority. For example, our study included 1,774 routine coagulation specimens on the day shift, 50% of which were completed by 122 minutes; half of the 4,206 day shift stat basic-7 specimens were completed within minutes. For only those tests in our study, 45% (19,259/42,414) were processed on a stat basis. Median turnaround time is 1.70 times faster for stat than for routine tests, although this value is heavily biased by chemistry tests and the day shift. We complete half of all stat tests within 44 minutes and half of all routine tests in 119 minutes. It appears our clinical laboratory performs an inordinately large percentage of the "basic-7" chemistry tests on a stat basis (Table 1). Our emergency and perinatal laboratories perform 25% (6,187/25,171) of all electrolyte analyses. The vast majority (97%) of analyses in these satellite laboratories are performed stat (Table 3). Our laboratory does not limit the number of stat tests by imposing a fee differential for stat testing. Results for tests requested on the evening and night shifts are completed more quickly than those on the day shift, within each priority class. Although staffing is considerably lighter on evening and night shifts, testing during

3 Vol.91 -No. 3 BRIEF SIENTIFI REPORTS 333 these hours is generally restricted to those analytes performed by automated instrumentation, with some exceptions (e.g., toxicology, body fluid cell counts, urinalysis [available on day and evening shifts only], Gram's stain [available on day and evening shifts only, with few exceptions], and blood bank services). The percentage of tests ordered stat increases from the day to the evening to the night shift, with approximately 97% of all studied analytes ordered stat on the night shift. We transmit or phone all inpatient stat results to the hospital ward or to the ordering physician. Our laboratory does not guarantee a turnaround time for most stat tests; however, we do expedite stat specimens through the laboratory and prioritize them ahead of the routine work load. Turnaround times are generally faster for those tests ordered stat, with considerable differences most notably on the day shift when the work load is the greatest (Table 2). The largest differences occur in the high-volume main hematology and chemistry laboratories. Although differences between stat and routine requests exist for most test groups, on the night shift the absolute minute differences tend to be small. In a few situations, stat tests appear to require more time than routine tests. These occur on the night shift when the number of actual routine tests is small (Table 2); the differences are only a few minutes. The difference in hematology clinic turnaround times (Table 3) most likely results from the small number of stat tests requested. The stat tests may have been ordered when a large number of requests were simultaneously received, with the most urgent requests prioritized by listing them as stat. Discussion Table 1. Analyte Grouping and Number of Assays per Analyte ategory/analyte oagulation APTT PT Fibrinogen FDP Protamine S0 4 Thrombin time Total coagulation Hematology B Platelet count Total hematology hemistry Basic 7 Sodium Potassium hloride Total 0 2 Serum urea nitrogen reatinine Glucose Total hem Basic 7 hemistry enzymes Amylase AST reatine kinase Total chemistry enzymes hemistry other Ammonia alcium, total alcium, ionized Magnesium Bilirubin, total Bilirubin, conjugated Total chemistry other Totals Routine 1,015 1, ,202 3,486 1,802 5,288 1,906 1,977 1,976 1,931 1,703 1,835 1,153 12, ,256 23,155 Stat ,512 1, ,984 1,975 2,422 2,084 2,034 1,404 1,4 1,7 12, ,500 19,259 Total 1,711 1, ,714 4,528 2,744 7,272 3,881 4,399 4,060 3,965 3,107 3,219 2,540 25, , , , ,756 42,414 The clinical laboratory plays a central role in the diagnosis and treatment of hospitalized and nonhospitalized patients. Quality control and quality assurance programs have been developed to assure the reliability and usefulness of the laboratory result. 6 ' 710 " Laboratory quality assurance programs for clinical pathology disciplines have become quite sophisticated. The ability of interlaboratory proficiency programs to improve the accuracy and precision of laboratory results has been well documented. 8 " /nterlaboratory proficiency results correlate positively with the percentage of American Society of linical Pathologists certified laboratory personnel 12 and with laboratory director qualifications. 9 Recent reports have called for a similar rigorous proficiency testing program for anatomic pathology disciplines. For a clinical laboratory test to be useful to the clinician, results should be available in a timely fashion. Few studies have examined turnaround times, and those reports in the literature have limited their discussions to stat testing. Barnett and associates 1 surveyed 53 hospitals from which they received usable responses. Respondents were questioned regarding which tests they offered on a stat basis, the stat turnaround time from specimen collection to completion, and any restrictions on the ordering or processing of stat requests in their respective institutions. Results of this survey are expectations for stat turnaround times and not actual laboratory performances. Diamond has reviewed the clinical reasons physicians use the laboratory. 2 Major reasons include diagnostic help, patient management, and establishment of patient baseline data and for defensive practice in our litigious society. Diamond comments that to be useful, particularly for the first two reasons, results must be reported in a timely

4 334 HILBORNE ET AL. Table 2. Turnaround Time (TAT) for Specimens by, Shift, and Priority A.J..P. March 1989 Routine Stat Shift n Median TAT (minutes) n Median TAT (minutes) Stat Percent Faster Percent Stat Tests oagulation 1, Hematology 4, hemistry Basic-7 10,923 1, ,206 4,0 4, hemistry Enzyme hemistry Other 1, , N For abbreviations, see Table I. fashion and laboratory performance is judged by many of our clinical colleagues in terms of timeliness more than by accuracy and precision. Watts reviews the medical relevance of laboratory tests and recommends monitoring the timeliness of result reporting. 18 Many clinical laboratories now have sophisticated LISs with the ability to track theflowof specimens from either the collection point or laboratory receipt to test completion and reporting. Westgard and Klee suggest turnaround time can be evaluated like any other quality control variable. 20 They comment that turnaround time may also be used to identify delayed or lost specimens. Our LIS routinely prints an incomplete specimen list for this purpose. Our study confirms turnaround time can be easily studied to measure laboratory output on a regular basis with the use of hardware and software that is already present in most laboratories. Stat tests represent a disruption to the normal laboratory work flow because such requests must be separately (A 3 c.2 B «S (A 0) *» oagulation Hematology Basle 7 Enzymes ham-other 0 s D B Table 3. Turnaround Time (TAT) for Specimens by Work Area and Priority Number of Tests (%).2 5 a> 2 Work Area Routine Stat Routine laboratory Satellite laboratories Stat laboratories linic laboratories 22,370 (66%) 221 (3%) 564 (94%) 11,717(34%) 7,505 (97%) 37 (6%) FIG. oagulation Hematology Basic 7 Enzymes hem-other 1. Graphic representation of median laboratory turnaround time by test group and shift for routine (A) and stat (B) tests.

5 Vol.91 -No. 3 BRIEF SIENTIFI REPORTS 335 identified and expedited through specimen-processing procedures. Singling out individual specimens decreases laboratory efficiency. Under prospective payment for hospital reimbursement, the laboratory is changing from a revenue center to a cost center; laboratory directors are facing increased administrative pressures to reduce operating costs. Special handling required by stat requests will, if in sufficient numbers, require additional laboratory personnel, increasing laboratory variable costs. We believe it is important to examine turnaround times for routine as well as stat specimens. One should measure turnaround times by shift and work area. In situations for which there is little difference between stat and routine turnaround times, we recommend laboratory directors evaluate the utility of processing all tests as they arrive, regardless of test priority. For example, during evening hours our laboratory completes routine specimens in a time similar to that of daytime stat requests. Physicians who order routine tests have implicit or perhaps explicit expectations regarding how long routinely requested specimens should take. If the laboratory is unable to efficiently result routine requests, more stat requisitions will be made, placing increased stress on laboratory management and technical personnel. This change in ordering pattern will slow all test results, including those stat tests that are truly ordered for medical emergencies. Analytic goals have been well defined for technical proficiency in the clinical laboratory. 5-7 ' 517 "' 9 Most facilities use rigorous guidelines to assure the precision and accuracy of laboratory results. We believe appropriate analytic goals for specimen turnaround time also need to be developed based on medical necessity and clinician expectation. In order to begin to define analytic goals for turnaround time, we must first have in place a mechanism whereby we can continually assess and reevaluate our performance in this important area. To our knowledge, this study is thefirstthat uses readily available electronic media to begin to develop such goals. References 1. Barnett RN, Mclver DD, Gorton WL: The medical usefulness of stat tests. Am J lin Pathol 1978;69: Diamond I: The clinical purposes of laboratory testing. Arch Pathol Lab Med 1988;112: Duckworth JK: Proficiency testing: its role in a voluntary clinical laboratory accreditation program. Arch Pathol Lab Med 1988; 112: Edinger SE: The Medicare and linical Laboratories Improvement Act of 1967 proficiency testing requirements and its relationship to the private sector. Arch Pathol Lab Med 1988;112: Fraser G: Desirable standards for hematology tests: a proposal. Am J lin Pathol 1987;88: Grannis GF, Grumer HD, Lott JA, Edison JA, Mcabe W: Proficiency evaluation of clinical chemistry laboratories. lin hem 1972;18: Harris EK: Statistical principles underlying analytic goal-setting in clinical chemistry. Am J lin Pathol 19;72: Hoeltge GA, Duckworth JK: Review of proficiency testing performance of laboratories accredited by the ollege of American Pathologists. Arch Pathol Lab Med 1987;111: Howanitz PJ: Use of proficiency test performance to determine clinical laboratory director qualifications. Arch Pathol Lab Med 1988;112: Ladenson JH: Patients as their own controls: use of the computer to identify "laboratory error." lin hem 1975;21: Laessig RH, Ehrmeyer SS: Proficiency testing programs promises, progress and problems. Arch Pathol Lab Med 1988;112: Lunz ME, astleberry BM, James K, Stahl J: The impact of the quality of laboratory staff on the accuracy of laboratory results. JAMA 1987;2: Nosanchuk JS, Gottmann AW: UMS and delta checks. A systematic approach to quality control. Am J lin Pathol I974;62: Ross JW: Evolution of evaluation criteria in the ollege of American Pathologists surveys. Arch Pathol Lab Med 1988;112: Ross JW: Goals for allowable analytic error better based on medical usefulness criteria. Am J lin Pathol 1986;85: Sheiner LB, Wheeler LA, Moore JK: The performance of delta check methods. lin hem 19;25: Skendzel LP, Barnett RN, Piatt R: Medically useful criteria for analytic performance of laboratory tests. Am J lin Pathol 1985,83: Watts NB: Medical relevance of laboratory tests. Arch Pathol Lab Med 1988;112: Westgard JO, Barry PL, Hunt MR, Groth T: A multi-rule Shewhart chart for quality control in clinical chemistry. lin hem 1981 ;27: Westgard JO, Klee GG: Quality assurance. In: Tietz NW. Textbook of clinical chemistry. Philadelphia: WB Saunders, 1986:424-4.