TY - JOUR
AU - Weiss, Ronald, L
AB - Abstract Background: Our laboratory, a large, commercial, esoteric reference laboratory, sought some form of total laboratory automation to keep pace with rapid growth of specimen volumes as well as to meet competitive demands for cost reduction and improved turnaround time. Methods: We conducted a systematic evaluation of our needs, which led to the development of a plan to implement an automated transport and sorting system. We systematically analyzed and studied our specimen containers, test submission requirements and temperatures, and the workflow and movement of people, specimens, and information throughout the laboratory. We performed an intricate timing study that identified bottlenecks in our manual handling processes. We also evaluated various automation options. Results: The automation alternative viewed to best meet our needs was a transport and sorting system from MDS AutoLab. Our comprehensive plan also included a new standardized transport tube; a centralized automated core laboratory for higher volume tests; a new “automation-friendly” software system for order entry, tracking, and process control; a complete reengineering of our order-entry, handling, and tracking processes; and remodeling of our laboratory facility and specimen processing area. Conclusions: The scope of this project and its potential impact on overall laboratory operations and performance justified the extensive time we invested (nearly 4 years) in a systematic approach to the evaluation, design, and planning of this project. Our laboratory is a commercial esoteric reference laboratory performing >2000 different procedures with an average daily volume of >18 000 accessions. We serve hospital clinical laboratories and other reference laboratories in all 50 of the United States, as well as several other countries, and receive mostly secondary specimens that have already been processed. We have averaged a 20% increase in specimen volume annually for most of our 18 years of existence. This continual high growth rate and the competitive cost pressures inherent in the laboratory environment led us to consider the possibility of some form of total laboratory automation (TLA). 1 However, automation in a setting that has little routine clinical laboratory testing and in which 85% of all incoming specimens are either refrigerated or frozen presented major challenges. Virtually all previous TLA endeavors have involved ambient specimens for routine chemistry, hematology, coagulation, and immunoassay tests, which have typically constituted a high proportion of the total workload of the laboratories implementing TLA (1)(2). In hospital laboratories and reference laboratories that serve physician offices, the majority of the laboratory testing is concentrated on routine testing, and 35–55 different tests may typically constitute 80% of such a laboratory’s volume, making such laboratories more amenable to TLA if volumes are sufficiently high. There are >200 TLA installations in laboratories in Japan and North America [Ref. (3), and B. Werner, Labotix Automation, Inc., personal communication]. Most of these systems include several or all of the following: (a) front-end processing to identify a specimen with a barcode and to match specimens to ordered tests; (b) gross sorting (e.g., separation of chemistry from hematology); (c) centrifugation; (d) aliquoting into barcode-labeled tubes; (e) direct sampling by analyzers that are interfaced to the automation system; and (f) specimen archive or storage. An excellent review of typical TLA elements was written by Bauer and Teplitz (4). The largely esoteric testing menu of our laboratory creates a different volume profile from that of other laboratories that have implemented TLA. Our top 50 tests comprise only 37% of our total volume, and the average daily volumes of these tests range from 50 to 500 samples. Our top 10 clinical laboratory tests might not appear on the top 10 lists or perhaps even the top 30 lists of other laboratories that have implemented TLA. To reach 80% of our volume requires >1000 different tests, and for tests at this level (1000 tests deep into our menu arranged by volume), we may receive an average of only 3–4 specimens per week for each different test. Thus, for an automation system to handle 80% of our volume, it would have to address the sorting of >1000 different tests, many of which have relatively few specimens each day. Each of the >1000 different tests that comprise the remaining 20% of our volume average three specimens per week or less. Typically, these low-volume tests are handled by each of the >100 employees in the Specimen Processing section once a year or less, but each employee is still expected to know exactly what to do with those requests. These intimidating challenges created a setting in 1995 in which the possibility of automation, although viewed as necessary because of our anticipated continuing high growth rate, was still only a glimmer on a distant horizon. This report details the methods we used to evaluate our needs relative to the possibility of automation. We determined the feasibility of automation in our laboratory and what it might accomplish. We review our systematic studies of temperature requirements, workflow, order-entry processes, specimen handling, and even the ergonomics of a specimen-processing workstation. We discuss our consideration of various alternatives for automation, leading to a plan for overall implementation that was focused not on automation per se, but on a complete reengineering of most of our preanalytic and postanalytic processes. Although some elements of our automation initiative have been described previously (5), that report lacked the details of our systematic evaluations, analyses, and planning that spanned nearly 4 years before our “go live” date in November 1998. Materials and Methods evaluation of temperature requirements We performed a detailed review of the temperature requirements for all tests that might potentially be part of an automation endeavor. Because we have clients in all 50 states and a few foreign countries and because our testing is largely esoteric, the majority of the specimens are shipped either frozen or refrigerated, with only a small percentage shipped ambient. In addition, some tests had been published in our laboratory’s users guide as “critical frozen” or “critical refrigerated”. Our analysis, therefore, included a study of our definition of “critical”, the volume of arriving specimens classified as critical frozen or critical refrigerated, and the impact of those volumes on our consideration of a possible automation solution. conceptual feasibility In January 1995, we learned of the AutoLabTM system developed by MDS Laboratories (Toronto, Canada), which used high-speed sorting machines, a unique standardized tube carrier (STC) that could be combined with other STCs to form test tube “racks”, and extensive process control software. A project team from our laboratory visited the MDS automated laboratory in April 1995, and over the next several months, feasibility discussions were held, which culminated in a consulting agreement with MDS that was signed in early 1996. This consulting agreement covered MDS’s assistance in our work flow analysis and in other aspects of evaluation of alternatives and planning, and ultimately it led to the design of an automation system. We had long recognized that a major issue for our laboratory was the sorting and repeated handling of small numbers of >2000 different tests. Moreover, we knew that a potential automation solution for our laboratory would likely not require centrifugation or aliquoting, although TLA systems being installed in high-volume laboratories in the mid-1990s usually had those features. Often, the sorting capabilities of these systems were limited to diverting specimens into major groups of tests such as chemistry, hematology, coagulation, and immunoassay, with each major group going into different processing or analytical lines that were part of the total system. Therefore, our feasibility evaluation focused on identifying systems that had strong sorting capabilities, not just centrifugation, aliquoting, or direct-from-track robotic sampling into analyzers. workflow analysis Fundamental to any decision to automate was an understanding of what processes were to be reengineered or automated and what efficiencies might be realized. A team of our managers and MDS AutoLab consultants met daily for 1 week to map the workflow at our laboratory. This map consisted of a diagram of the movement of all entities (e.g., tubes, cups, bottles, racks, shipping containers, requisitions, reports, lists, and labels) in the building and the people or sections who handled those entities. The map included every activity from the arrival of work at the exterior door of the laboratory until the completion of testing, including archival storage of specimens for possible repeat or additional testing. timing study The workflow map was then used to design a detailed timing study in which all of the steps in the process were timed from the arrival of shipments to the completion of specimen storage after test result verification. The purpose of this study was to determine the average elapsed times for the various steps in the process to identify those steps that should be a focus of our reengineering and/or automation and, in particular, to identify bottlenecks that hampered timely movement of the work. In the study, each shipping container was stickered with a tracking slip noting the container number and its time of arrival. Then, when the container was unpacked, the time was noted again and the bags of specimens within were marked with new tracking slips linked to the container number. Approximately 10% of all specimens were individually tracked through all steps in the process, including order entry, sorting, transfer to the laboratory sections, work listing, testing, return to storage, and storage. This tracking of process times was performed for a full 24 h to cover all courier delivery cycles of locally originating specimens and shipments arriving from the airport. needs assessment On the basis of the workflow analysis, the timing study, and the temperature study, we performed a specific analysis to identify our needs and areas that required possible process reengineering. Needs assessment and careful formulation of goals have been emphasized by other authors as critical steps in planning for TLA (6)(7). outside consultants Subsequent to our evaluations of workflow and needs, while we were considering our alternatives and developing our overall plan, we used an outside consulting firm to independently assess our needs and plans. Two engineers from Argent Consulting Services, Inc. (Oklahoma City, OK) spent ∼3 weeks specifically studying our Specimen Processing and Automated Core Laboratory sections, the latter having just been formed. evaluation of alternatives, including automation Various alternative approaches to improving our workflow were considered. In addition to an intention to consolidate some of our higher volume testing into an automated core laboratory, we considered the possibility of reengineering our basic processes, but without implementing any automation. We also considered several alternatives for automation, including both track (conveyor)-based systems and other systems. The nonconveyor alternatives we considered included automated guided vehicles (AGVs) to deliver specimens to laboratory sections, equipping an AGV with a robotic arm and a barcode reader to automate unloading of specific tubes at specific locations, and the use of stand-alone automated sorters, such as the AutoLab AutoSortsTM. The track (conveyor) systems made by Lab-Interlink®, Coulter®, Boehringer Mannheim Hitachi CLASTM, and MDS AutoLab were all considered. 2 use of project planning Microsoft Project© for Windows 95, Ver. 4.1, was used to track and plan all aspects of this complex project, including coordinating the facilities renovation; the development and testing of Expert Specimen Processing (ESP); the design, construction, factory testing, shipment, installation, and on-site testing of the automation system; the training of employees; and numerous other essential steps. Results evaluation of temperature requirements Historically, some sections had used the term “critical” (e.g., critical frozen) if the analyte would begin to deteriorate as soon as the specimen was thawed or warmed, whereas other sections had used the term critical out of concern for the reliability of the courier transport. In a meeting with laboratory section supervisors and managers, we arrived at a common definition of critical, which was the former one above. Each laboratory was then provided a spreadsheet to review their tests and temperature requirements and return with corrections. On the basis of the volumes of each test, this common definition for critical reduced the total number of critical specimens arriving at the laboratory from 20% to 10% of the total, an important step because these specimens would likely not be transported on an automation system. The temperature evaluation also provided information on the type of specimen container required for each test. Approximately 10% of all arriving specimens were odd-sized containers (e.g., cultures, probes, kidney stones, fecal specimens, and urine clinical toxicology screens). These 10% could clearly not be transported by a tube-based automation system. Added to the 10% with a critical temperature designation, this meant that only 20% of our total arriving specimens were potentially not transportable. Assuming that provisions could be made to assure the temperature requirements of the potentially transportable 80%, an automation system appeared to be a worthwhile objective. workflow analysis The workflow map developed by the joint ARUP-MDS project team is shown in Fig. 1. Table 1 lists the codes for the physical entities (things such as tubes, racks, and requisitions) that are mapped in Fig. 1. The work flow map revealed that, not counting the actual test analysis, each specimen was handled on average by at least 10 individuals, each performing a different step in the process. Some specimens required even more handling, such as a urine specimen being checked for pH or thawed to prepare aliquots. Four sorting steps were identified as present for nearly every specimen: (a) sorting by Specimen Processing personnel into laboratory section groups after completion of order entry; (b) sorting by laboratory section personnel into work center groups, including reading of the barcoded accession number to record in the laboratory information system (LIS) the transfer of the specimen from Specimen Processing to the laboratory section; (c) sorting by work center personnel into individual tests and building work lists in the LIS; and (d) sorting by laboratory personnel after completion of the test by the time and temperature to be used for archival storage. We concluded that a major focus of our reengineering and automation endeavor should be to reduce this excessive sorting and handling, including if possible, the handling to update the computer tracking and to build work lists. Figure 1. Open in new tabDownload slide Map of the workflow at ARUP before implementation of the automation system. Panels A, B, and C align left to right in that order. Each entity code (E1–E15) that appears in Fig. 1 is defined in Table 1. See text for explanation and discussion. SAM, Substance Analysis and Management (ARUP’s acronym for employment and forensic drug testing); CYTO, cytology; QA, quality assurance; Acc#, accession number; ID, identification. Figure 1. Open in new tabDownload slide Map of the workflow at ARUP before implementation of the automation system. Panels A, B, and C align left to right in that order. Each entity code (E1–E15) that appears in Fig. 1 is defined in Table 1. See text for explanation and discussion. SAM, Substance Analysis and Management (ARUP’s acronym for employment and forensic drug testing); CYTO, cytology; QA, quality assurance; Acc#, accession number; ID, identification. Table 1. Entity codes for the various entities in Fig. 1. Code . Entity . E1 Shipping container E2 Specimen bag E3 Requisition or manual work list E4 Specimen E5E Aliquot tube (empty) E5F Aliquot tube (not empty) E6 Prelabeling bin (E2 + E8) E7 Postlabeling bin (E5F + E10 + E11) E8 Labels E9 Exception form E10 Labeled specimen E11 Specimen bundle (banded E10 + 1 or > E5E) E12 Rack E13 Analytical work list E14 Storage box E15 Barcoded e-mail notice of add test Code . Entity . E1 Shipping container E2 Specimen bag E3 Requisition or manual work list E4 Specimen E5E Aliquot tube (empty) E5F Aliquot tube (not empty) E6 Prelabeling bin (E2 + E8) E7 Postlabeling bin (E5F + E10 + E11) E8 Labels E9 Exception form E10 Labeled specimen E11 Specimen bundle (banded E10 + 1 or > E5E) E12 Rack E13 Analytical work list E14 Storage box E15 Barcoded e-mail notice of add test Open in new tab Table 1. Entity codes for the various entities in Fig. 1. Code . Entity . E1 Shipping container E2 Specimen bag E3 Requisition or manual work list E4 Specimen E5E Aliquot tube (empty) E5F Aliquot tube (not empty) E6 Prelabeling bin (E2 + E8) E7 Postlabeling bin (E5F + E10 + E11) E8 Labels E9 Exception form E10 Labeled specimen E11 Specimen bundle (banded E10 + 1 or > E5E) E12 Rack E13 Analytical work list E14 Storage box E15 Barcoded e-mail notice of add test Code . Entity . E1 Shipping container E2 Specimen bag E3 Requisition or manual work list E4 Specimen E5E Aliquot tube (empty) E5F Aliquot tube (not empty) E6 Prelabeling bin (E2 + E8) E7 Postlabeling bin (E5F + E10 + E11) E8 Labels E9 Exception form E10 Labeled specimen E11 Specimen bundle (banded E10 + 1 or > E5E) E12 Rack E13 Analytical work list E14 Storage box E15 Barcoded e-mail notice of add test Open in new tab timing study The timing study was conducted for a 24-h period on a typically busy day (Wednesday) on which ∼5000 total specimens arrived. The data measurements for some activities extended beyond 24 h. An extensive spreadsheet analysis enabled us to produce a summary report with median times and ranges for each step of the process for each courier delivery cycle. At the time of this study, 70% of incoming specimens arrived between 2200 in the evening and 0700 in the morning, reflecting the time for air transport from our client locations. This analysis measured the flow and timing at various steps for each cycle of courier pick-up of shipments at the airport (such as at 2200, 0000, 0200, 0400, 0700, and so forth). For each of these cycles we determined the waiting times at various steps in our “assembly line” process as well as the total time from arrival until pick-up of processed specimens by the laboratory sections. The analysis was extremely helpful in developing an automation plan focused on transport and sorting because it revealed the steps that were bottlenecks or unacceptable time wasters. These steps included the waiting time from courier manifesting to order entry, the manual transfers through the Specimen Processing production line (which resembled a factory assembly line), the waiting time from the final sort in Specimen Processing to pick-up by the laboratory sections, and the waiting time from completion of testing until completion of storage. outside consultants The recommendations of the independent consultants included (a) changing the assembly line in Specimen Processing to individual workstations; (b) standardizing processing procedures with a “best practice” solution that required fewer hand-offs, increased accountability, decreased turnaround times, and improved quality; (c) relocating all high-volume and automated testing to the Automated Core Laboratory; (d) positioning the Automated Core Laboratory as close to Specimen Processing as possible; and (e) making the Automated Core Laboratory the first stop for any specimens with tests to be performed in that section. In summary, the independent consultant’s detailed report both supported our plan and provided additional helpful recommendations. needs assessment We identified a list of eight key needs based on the work flow analysis, timing study, and consultant recommendations. These were identified as areas for process improvement through reengineering and/or automation and are listed in Table 2A. To the extent that we could respond to these needs, we would eliminate many of the bottlenecks and much of the unnecessary handling of specimens identified by the workflow map. Concurrent with identifying our needs, we developed the concept of a new software system, ESP (see a subsequent section in this report). As this software system became part of our overall plan and was being designed in a broad sense, eight additional needs were identified, which are listed in Table 2B and which addressed specific problems identified in our workflow map. Table 2. Identified needs. A. Needs for process reengineering and/or automation identified by workflow analysis  • To reduce the total number of times that individual tubes are handled before and after analytical testing  • To reduce the number of separate sorting steps, a subset of the first need  • To deliver specimens closer to the laboratory areas instead of to a central refrigerator and freezer in Specimen Processing, and no longer depend on the individual laboratories to pick up their own specimens on their own schedules  • To maximize the number of specimens available at the start of all test runs  • To convert Specimen Processing from an assembly line to individual workstations to improve efficiency, quality, and accountability  • To eliminate “running around” to find shared specimens and, as a result, reduce lost specimens and tests not performed  • To consolidate higher volume, automated testing in a centralized Automated Core Laboratory  • To maximize our scarce human resources by better matching job assignments to skills B. Additional process needs identified as part of the ESP project  • To develop a rules-based order-entry or accessioning system, which would guide Specimen Processors to order tests and label tubes with greater accuracy  • To reduce training time to achieve a minimal degree of competency for Specimen Processors, who typically have a higher than average turnover rate  • To eliminate manual barcode reading to change the LIS order status of a given accession number from “Central Collect” status to “In Laboratory” status  • As a subset of the item above, to have the order status update occur in real time relative to the actual transfer of the tubes. In the past, the laboratories often did not change the order status to “In Laboratory” until they were ready to build a worksheet although the tubes might have been in their possession for hours or even days. This meant that the LIS might still indicate a tube’s location as Specimen Processing where the tube was in the laboratory section  • To change the level of this order status update from accession number level to test level  • To build more worksheets dynamically instead of in accession number order and to automate dynamic worksheet building  • To automate the existing specimen archive process, which required manual reading of every tube’s barcode number  • To know with greater accuracy the exact location of every tube at any point in time in the laboratory A. Needs for process reengineering and/or automation identified by workflow analysis  • To reduce the total number of times that individual tubes are handled before and after analytical testing  • To reduce the number of separate sorting steps, a subset of the first need  • To deliver specimens closer to the laboratory areas instead of to a central refrigerator and freezer in Specimen Processing, and no longer depend on the individual laboratories to pick up their own specimens on their own schedules  • To maximize the number of specimens available at the start of all test runs  • To convert Specimen Processing from an assembly line to individual workstations to improve efficiency, quality, and accountability  • To eliminate “running around” to find shared specimens and, as a result, reduce lost specimens and tests not performed  • To consolidate higher volume, automated testing in a centralized Automated Core Laboratory  • To maximize our scarce human resources by better matching job assignments to skills B. Additional process needs identified as part of the ESP project  • To develop a rules-based order-entry or accessioning system, which would guide Specimen Processors to order tests and label tubes with greater accuracy  • To reduce training time to achieve a minimal degree of competency for Specimen Processors, who typically have a higher than average turnover rate  • To eliminate manual barcode reading to change the LIS order status of a given accession number from “Central Collect” status to “In Laboratory” status  • As a subset of the item above, to have the order status update occur in real time relative to the actual transfer of the tubes. In the past, the laboratories often did not change the order status to “In Laboratory” until they were ready to build a worksheet although the tubes might have been in their possession for hours or even days. This meant that the LIS might still indicate a tube’s location as Specimen Processing where the tube was in the laboratory section  • To change the level of this order status update from accession number level to test level  • To build more worksheets dynamically instead of in accession number order and to automate dynamic worksheet building  • To automate the existing specimen archive process, which required manual reading of every tube’s barcode number  • To know with greater accuracy the exact location of every tube at any point in time in the laboratory Open in new tab Table 2. Identified needs. A. Needs for process reengineering and/or automation identified by workflow analysis  • To reduce the total number of times that individual tubes are handled before and after analytical testing  • To reduce the number of separate sorting steps, a subset of the first need  • To deliver specimens closer to the laboratory areas instead of to a central refrigerator and freezer in Specimen Processing, and no longer depend on the individual laboratories to pick up their own specimens on their own schedules  • To maximize the number of specimens available at the start of all test runs  • To convert Specimen Processing from an assembly line to individual workstations to improve efficiency, quality, and accountability  • To eliminate “running around” to find shared specimens and, as a result, reduce lost specimens and tests not performed  • To consolidate higher volume, automated testing in a centralized Automated Core Laboratory  • To maximize our scarce human resources by better matching job assignments to skills B. Additional process needs identified as part of the ESP project  • To develop a rules-based order-entry or accessioning system, which would guide Specimen Processors to order tests and label tubes with greater accuracy  • To reduce training time to achieve a minimal degree of competency for Specimen Processors, who typically have a higher than average turnover rate  • To eliminate manual barcode reading to change the LIS order status of a given accession number from “Central Collect” status to “In Laboratory” status  • As a subset of the item above, to have the order status update occur in real time relative to the actual transfer of the tubes. In the past, the laboratories often did not change the order status to “In Laboratory” until they were ready to build a worksheet although the tubes might have been in their possession for hours or even days. This meant that the LIS might still indicate a tube’s location as Specimen Processing where the tube was in the laboratory section  • To change the level of this order status update from accession number level to test level  • To build more worksheets dynamically instead of in accession number order and to automate dynamic worksheet building  • To automate the existing specimen archive process, which required manual reading of every tube’s barcode number  • To know with greater accuracy the exact location of every tube at any point in time in the laboratory A. Needs for process reengineering and/or automation identified by workflow analysis  • To reduce the total number of times that individual tubes are handled before and after analytical testing  • To reduce the number of separate sorting steps, a subset of the first need  • To deliver specimens closer to the laboratory areas instead of to a central refrigerator and freezer in Specimen Processing, and no longer depend on the individual laboratories to pick up their own specimens on their own schedules  • To maximize the number of specimens available at the start of all test runs  • To convert Specimen Processing from an assembly line to individual workstations to improve efficiency, quality, and accountability  • To eliminate “running around” to find shared specimens and, as a result, reduce lost specimens and tests not performed  • To consolidate higher volume, automated testing in a centralized Automated Core Laboratory  • To maximize our scarce human resources by better matching job assignments to skills B. Additional process needs identified as part of the ESP project  • To develop a rules-based order-entry or accessioning system, which would guide Specimen Processors to order tests and label tubes with greater accuracy  • To reduce training time to achieve a minimal degree of competency for Specimen Processors, who typically have a higher than average turnover rate  • To eliminate manual barcode reading to change the LIS order status of a given accession number from “Central Collect” status to “In Laboratory” status  • As a subset of the item above, to have the order status update occur in real time relative to the actual transfer of the tubes. In the past, the laboratories often did not change the order status to “In Laboratory” until they were ready to build a worksheet although the tubes might have been in their possession for hours or even days. This meant that the LIS might still indicate a tube’s location as Specimen Processing where the tube was in the laboratory section  • To change the level of this order status update from accession number level to test level  • To build more worksheets dynamically instead of in accession number order and to automate dynamic worksheet building  • To automate the existing specimen archive process, which required manual reading of every tube’s barcode number  • To know with greater accuracy the exact location of every tube at any point in time in the laboratory Open in new tab Our evaluations also identified the need for process control in the automation plan. We knew that many of our employees spent an inordinate amount of time searching for “shared” specimens (tubes needed for two or more tests performed in different laboratory sections). We needed a system that would not only transport tubes expeditiously, but had increased tracking capabilities. We also knew that our laboratory sections retained their specimens for up to 2 days after completion of testing because they did not trust the speed of our manual specimen storage system. We thus defined a need for storage to occur promptly so that the specimens would be available for repeat or additional testing. evaluation of alternatives, including automation In our setting of high total volumes, but with relatively low volumes of many individual tests, the AGV alternative was considered by our project team to be unworkable. The AGVs would likely have had to be manually loaded and unloaded, and all of the sorting steps identified in the work flow map would still exist. Automated loading and unloading of AGVs is possible, but to a relatively limited number of sort groups. In short, the AGV accomplished only transport without eliminating manual sorting. The project team also believed that use of a robotic device for sorting and loading an AGV could have been a potential bottleneck because of our large volumes, rather than a labor and time saver. The alternative of an AGV equipped with a robotic arm to unload specimens at laboratory locations was viewed as unacceptable because of slow throughput, and the development of standardized unloading stations for more than 15 laboratory sections was also viewed as problematic. We also considered the use of stand-alone high-speed sorters such as those made by AutoLab, but felt that they would have led to lines of laboratory assistants waiting to sort their laboratory section’s specimens, with resulting delays and congestion. Specimen Processing would still have had to sort specimens by laboratory section, and the laboratory assistants would still have had to come to a central freezer and refrigerator to pick up their tubes. The track (conveyor) systems made by Lab-Interlink, Coulter, Boehringer Mannheim Hitachi CLAS, and MDS AutoLab were all considered. The Lab-Interlink and Coulter systems were not selected for similar reasons. The specimen carriers used by those systems were uniquely shaped and served only for use while on the conveyor system. These carriers could not be effectively used away from the conveyor, thus requiring that the specimen tubes had to be unloaded to other racks or carriers. Moreover, neither system had any sorting capability other than a left vs right at selected points on the track. The Hitachi CLAS system uses a five-position rack, requiring that five tubes for the same or similar tests be sorted together to go to a common destination (analyzer) for optimal efficiency. It is an unlikely occurrence at ARUP for five tubes for the same test to be seen together at the same time in Specimen Processing. Indeed, only 50 tests out of the >2000 tests performed even average five or more tubes arriving per hour, and these tubes are likely to be randomly distributed among the workstations in Specimen Processing. Compared with the MDS AutoLab system, sorting on the Hitachi CLAS system is organized for larger numbers of tubes for fewer sort groups. In general, all three of these systems were strongly oriented toward laboratories performing high volumes of routine chemistry and hematology tests, rather than the lower volumes of many tests characteristic of our laboratory. The MDS AutoLab system, on the other hand, had several key design features that made it attractive as a possible solution for our specific needs. The first feature was that the AutoLab system uses unique STCs, which when not on the conveyor system can be snapped together in two directions to make tube racks of any convenient size. This eliminates extra handling and the need to transfer tubes to other racks whether by people or robots. These racks can be transferred to a refrigerator or freezer or carried to laboratory destinations. The second feature was that the AutoSorts are each capable of sorting up to 1000 tubes/h into 30 different user-definable lanes of individual STCs, a much more useful design for ARUP’s purposes than a robot moving 1 tube at a time into 1 of 30 different racks (as an example of other types of sorting available). AutoSort lanes can be organized as individual tests or groups of related tests to best meet the needs of the laboratory. Finally, the computer system developed by AutoLab to support their system, AutoLab Process eXpert, or APXTM, had features that, in conjunction with ESP, offered the potential for our laboratory to meet many of our identified needs, especially those that would benefit from the process controller in the APX system. We were fortunate that the AutoLab team was well versed in process control, even having published on this subject (8), and that together, we were able to visualize how ESP could integrate with APX process control most effectively. overall plan for reengineering and automation The overall plan that resulted from our analysis of needs and the evaluation of alternatives had the following elements: Standardized specimen transfer tube. It was considered critical to the success of the automation initiative to use standardized transport tubes for many of the secondary specimens that constitute most of our work. The tube would need to fit the automation system STCs as well as the racks or carousels of as many automated analyzers and pipetting stations as possible throughout the laboratory. The nominal tube size would be 16 × 100 mm, the same as the most widely used evacuated blood collection tubes, except that we preferred slightly <16 mm in diameter so that it would fit freely into the STCs and analyzer racks or carousels even with as many as four different labels affixed. The tube would be of plastic with a tight-fitting screw cap to prevent leaks during air transport. A tube produced by Sarstedt (cat. no. 62.611.009) met these criteria. It is a false-bottom polypropylene tube, 15.3 × 92 mm. The tube was tested with as many as four labels affixed and was successfully spun in STCs for barcode reading. It was also tested under conditions of reduced pressure and did not leak. After its implementation, this tube replaced specialized tubes for three different automated analyzers in our Automated Core Laboratory as well as a 90-mL urine bottle used for most urine tests. We began using this tube nearly 2 years before implementation of the automation system, at which time we determined that ∼80% of all arriving tubes were this tube. This made our automation initiative more successful than if we had not attempted to have a standardized tube. A percentage >80% is not likely because we do not supply this tube to those clients who are reference laboratories themselves and because some tests require primary specimens or are otherwise not compatible with a false-bottom tube (e.g., whole-blood lead, our highest volume test). Moreover, some clients have preferred their own transfer tubes, and we planned that we would receive a certain percentage of nonstandardized tubes. Formation of an Automated Core Laboratory. The second element of our plan was to centralize in one laboratory section a substantial percentage of our higher volume testing that used automated analyzers. The testing to be relocated came from the General Chemistry, Endocrinology, Special Chemistry, and Immunology sections and accounted for ∼18% of our total volume. The remaining manual chemistry tests in the General Chemistry laboratory were relocated to other laboratory sections and this laboratory was discontinued as a section. In addition to volume, principal criteria for relocation to the Automated Core Laboratory were that the tests were nonbatch tests, did not require manual steps such as extractions, and could utilize existing automated analyzers. The criteria for relocating analyzers to the Automated Core Laboratory included compatibility with the ARUP standardized transfer tube, support of a bidirectional host query interface and autoverification, random access capability, and (preferably) no use of carousels for specimens. It was also part of our plan to increase over several years the percentage of our total testing that was performed in the Automated Core Laboratory and to reduce the number of different analyzer platforms in use. Reengineering of processes. Fundamental to our plan was the need to reengineer various processes. Although much reengineering would clearly result from automation, other major changes were independent of automation, such as the replacement of the Specimen Processing assembly line with individual workstations to gain greater efficiency, quality, and accountability. The development of the ESP rules-based order-entry system was expected to eliminate much of the guesswork and memorization of exceptions typically associated with the specimen-processing function. The combination of ESP and automation was expected to eliminate much manual sorting and handling of tubes as well as to dramatically change handling of specimens that must by shared by two or more laboratories. In addition, the automation system, by virtue of transporting processed specimens to high-speed sorters that were closer to the laboratory areas, would eliminate congestion at the centralized freezer and refrigerator where the Specimen Processing section had been placing processed specimens sorted by laboratory section. Finally, we planned to eliminate as much as 80–90% of the barcode reading associated with archiving finished specimens to storage. Development of ESP. Early in our evaluations, we determined that the container identification capability that could be added to our existing LIS (Cerner Pathnet, Ver. 3.06) would not adequately meet our needs for tracking individual tubes at the level we desired. We therefore elected to develop a new, Windows NT, server-based, “front-end” system for order entry, which would provide an independent container identification system and would interface between the Pathnet LIS and APX. This software development project also became an opportunity to develop a new rules-based accessioning system that would replace order entry in Pathnet. The considerable benefits of developing a new accessioning system led to the name for this new system, Expert Specimen Processing (ESP). These benefits included the ability to add modules and functionality as needed to meet our complex needs and having a defined rules set and an extensive database that would substantially reduce errors, inefficiencies, and training time in Specimen Processing. In our plan, ESP would be developed using Delphi (Inprise), a powerful object-oriented software development tool that uses the Pascal language and a graphical user interface. SQL Server (Microsoft) would be used as the relational database management system. In addition, a client/server model would be used on our local area network with the Windows NT operating system used on both the client and server ends. In our design, all accessioning would occur in ESP and would conclude with the printing of labels with two different barcode identifiers on each label, one for the accession number and one for a unique container tracking number. These barcodes would be of different symbologies so that the barcode readers on the transport system would not read the accession numbers, whereas the analyzers and hand-held barcode readers in the laboratory sections would not read the container tracking numbers. ESP would send the completed orders to Pathnet, which would function in every other way as previously. ESP would also send “orders” to APX, consisting of the container number followed by a sequence of one or more test codes ordered on the individual container. The test code sequence sent to APX would thus comprise a route for each tube on the transport system. However, the main attribute of the ESP Accessioning module’s design was to make it a rules-based system that would use a large database of all acceptable (not just preferred) specimen types. Unique in our design of ESP compared with most LIS or accessioning systems is that the numbers and types of tubes and specimens and their temperatures were to be entered for each order. ESP would then use this information to determine whether the correct specimens had been submitted for the ordered tests and, if not, to create an exception that would be transferred to our Exception Handling staff to process. ESP was also to make complicated decisions, for example, which specimens to aliquot, which tubes to place on the track, which tubes to route to multiple tests and in what sequence, and which tubes to send directly to storage because they were duplicates. Each specimen processor would use the ESP Accessioning module at one of 30 workstations equipped with a personal computer, 17-inch monitor, barcode reader, and barcode label printer. The workstations (Fig. 2) were designed by a task force of employees and reviewed by an ergonomics expert. Several designs and prototypes were evaluated before we selected an improved design that was installed. As more have been built for expansion or replacement, we have continually improved the design and quality of materials used, and we are currently on our sixth version. Figure 2. Open in new tabDownload slide Specimen processing workstation. Features include a personal computer equipped with laser barcode reader and barcode label printer, a hopper for standardized transfer tubes and caps, shelves for supplies, a washable Corian® work surface with a rim around the edges to prevent tubes from falling on the floor, styrofoam bins with dry ice for frozen specimens, and plastic bins with chilled gel packs for refrigerated specimens. Figure 2. Open in new tabDownload slide Specimen processing workstation. Features include a personal computer equipped with laser barcode reader and barcode label printer, a hopper for standardized transfer tubes and caps, shelves for supplies, a washable Corian® work surface with a rim around the edges to prevent tubes from falling on the floor, styrofoam bins with dry ice for frozen specimens, and plastic bins with chilled gel packs for refrigerated specimens. Approximately 83% of our incoming work is electronically preentered into Pathnet. Only 17% requires manual entry of the order information from paper requisitions. Our plan was to download electronic orders from Pathnet to ESP to await the arrival of the shipments. Specimen processors would then access lists of these pending orders in ESP and receive the orders for those tubes that were present by entering the temperatures, number of tubes, and types of specimens and then printing the barcode labels. In addition to the Accessioning module, we also planned that ESP would have modules for Inquiry, Storage, and Specimen Checkout. As tubes completed portions of their route (i.e., arrived at sorters), update messages would be sent from APX to ESP and ESP would update a tracking file for each tube, which could be accessed by client number, patient name, accession number, or container number through ESP Inquiry. Every specimen to be stored in our centralized storage and retrieval system would have a special test code in the ESP database to designate its storage time and temperature. All specimens would be stored at the same temperature as originally required for the test, but there would be no specific tracking of a tube’s temperature history. The storage test code would be the last code in the routing sequence sent from ESP to APX. A sorter, used exclusively for storage, would sort by storage codes into specified lanes and keep track of the barcode sequences in each lane. The ESP Storage module would transfer these sequences of barcode container numbers into a file of exact storage locations as the tubes were manually transferred to the storage boxes, thus saving ∼85% of the storage barcode reading of our previous manual method. For each sequence of 20 tubes, it would be necessary only to read the first and last tubes to identify the sequence being transferred, and, after transfer to the box, to read the barcode of a random tube highlighted on the screen for quality assurance of the transfer. The Specimen Checkout module would enable employees to “check out” specimens for repeat testing only after they provided their laboratory section and user identification code to the system. If they needed to examine several tubes from the same patient to find the correct tube, they could “uncheck” the unneeded tubes. Each specimen that was checked out would automatically send a new route to APX with the storage test code so that all checked out tubes could be returned to the track and automatically restored without a need for employees to segregate them from other tubes. Design of automated transport and sorting system. Although dependent on the successful implementation of ESP, the installation of an MDS AutoLab transport system with five high-speed sorting machines was clearly a critical element of our plan. The system jointly designed by ourselves and MDS AutoLab consisted of 360 feet of conveyor or track that could transport individual tubes in STCs. Robotic devices would perform mergers of two tracks into one, transfers between tracks, or diversions to branch tracks. Three “feeder” tracks would transport tubes from 30 Specimen Processing workstations to the main track, which was essentially a continuous loop. The main line would deliver specimens to four AutoSort sorters, each of which could sort up to 1000 tubes/h into 30 user-defined lanes. Three of the sorters were positioned to serve the laboratory areas nearest them, whereas the fourth sorter was for specimen storage. This design provided a total of 90 primary sort lanes in the three laboratory sorters compared with ∼35 manual sort groups used by Specimen Processing before automation. This meant that tubes would not be sorted just by laboratory section and temperature, but could be sorted into high-volume tests or groups of related tests. A fifth sorter, not connected to the track, was to serve as a backup sorter and for automated worksheet building when the latter was developed. In the design, when tubes arrive at each sorter, APX would send a message to ESP to update the tracking information. ESP, in turn, would send a similar message to Pathnet to change the order status of the arriving tube from “Central Collect” to “In Laboratory”. Automation of this update of order status would eliminate a major manual step in all laboratories. The speed of the track (only 4.5 min between the two most remote points on the system) would eliminate much congestion and human traffic in the Specimen Processing area. Although the speed would not allow frozen specimens to thaw, we nevertheless planned to still handle critical frozen specimens manually to protect labile analytes. It was believed that the transport speed combined with immediate sorting would make specimens available for testing several hours sooner than before automation. Status tracking in both ESP and Pathnet would be much more accurate and would reflect exactly a specimen’s location. Facility renovations. Substantial facility renovations were required to accommodate the desired objectives. A hallway separating Specimen Processing from the laboratory sections was removed, creating essentially a large open facility. Most of the automation system was located in this former hallway, as was the Specimen Processing expansion. An additional benefit was the fostering of more of a team atmosphere by not separating laboratory personnel from processing personnel. Other facility needs included provisions for compressed air for the robotics and a large uninterrupted power supply system to support the entire automation system and its computers. Discussion We believe the plan described in this report was a good plan with a high probability of success. The various evaluations performed by our team contributed immensely to a thorough understanding of our operations and needs. In particular, we note the temperature evaluation and the work flow map (Fig. 1) as essential to determining whether automation was feasible and which of our existing processes automation should address. The map provided a detailed picture of what actual operations were being performed and identified the steps that could possibly be eliminated through reengineering. Similar maps are contained in the previously cited chapter by Middleton and Mountain (8). We are strong advocates for the completion of such a map by any laboratory considering automation. The timeline for this total project is summarized in Table 3, including some of the postimplementation phases discussed in Part 2, a companion report to this one (9). Part 2 also describes our experiences and performance results over a 3-year period subsequent to implementation, including a discussion of what went right and what went wrong with our plan. Some of the elements reported here were planned in conjunction with our design of the AutoLab system but were implemented independently. We believe that these elements would have had positive impacts on our operations regardless of whether we implemented an automation system but were much more effective in combination with the track and ESP. These elements included the implementation of a standardized transport tube for submission of specimens, the formation of an Automated Core Laboratory, and the reengineering of Specimen Processing from an assembly line to individual workstations. The standardized tube was implemented 2 years before the implementation date for automation, at which time ∼80% of all arriving tubes were in this new tube. This reduced the need to transfer from other tubes for placement on the automation system or for analyzers that now had been adapted to the standardized tube. Table 3. Timeline of automation initiative. January 1995 ARUP learns of MDS AutoLab system April 1995 First visit by project team to MDS Laboratories (Toronto, Canada) June 1996 Temperature, workflow, and timing evaluations completed November 1996 Formation of Automated Core Laboratory February 1997 Introduction of standardized transfer tube October 1997 Implementation of single work stations in Specimen Processing section November 1997 AutoLab system design completed; contract signed April 1998 Introduction of version 1.0e of ESP Accessioning June 1998 AutoLab system installed November 17, 1998 Go live with AutoLab system (with version 2.1b of ESP Accessioning) May 1999 Key software upgrades to ESP and APX May 2000 New AutoSort sorters; 18 more Specimen Processing work stations January 1995 ARUP learns of MDS AutoLab system April 1995 First visit by project team to MDS Laboratories (Toronto, Canada) June 1996 Temperature, workflow, and timing evaluations completed November 1996 Formation of Automated Core Laboratory February 1997 Introduction of standardized transfer tube October 1997 Implementation of single work stations in Specimen Processing section November 1997 AutoLab system design completed; contract signed April 1998 Introduction of version 1.0e of ESP Accessioning June 1998 AutoLab system installed November 17, 1998 Go live with AutoLab system (with version 2.1b of ESP Accessioning) May 1999 Key software upgrades to ESP and APX May 2000 New AutoSort sorters; 18 more Specimen Processing work stations Open in new tab Table 3. Timeline of automation initiative. January 1995 ARUP learns of MDS AutoLab system April 1995 First visit by project team to MDS Laboratories (Toronto, Canada) June 1996 Temperature, workflow, and timing evaluations completed November 1996 Formation of Automated Core Laboratory February 1997 Introduction of standardized transfer tube October 1997 Implementation of single work stations in Specimen Processing section November 1997 AutoLab system design completed; contract signed April 1998 Introduction of version 1.0e of ESP Accessioning June 1998 AutoLab system installed November 17, 1998 Go live with AutoLab system (with version 2.1b of ESP Accessioning) May 1999 Key software upgrades to ESP and APX May 2000 New AutoSort sorters; 18 more Specimen Processing work stations January 1995 ARUP learns of MDS AutoLab system April 1995 First visit by project team to MDS Laboratories (Toronto, Canada) June 1996 Temperature, workflow, and timing evaluations completed November 1996 Formation of Automated Core Laboratory February 1997 Introduction of standardized transfer tube October 1997 Implementation of single work stations in Specimen Processing section November 1997 AutoLab system design completed; contract signed April 1998 Introduction of version 1.0e of ESP Accessioning June 1998 AutoLab system installed November 17, 1998 Go live with AutoLab system (with version 2.1b of ESP Accessioning) May 1999 Key software upgrades to ESP and APX May 2000 New AutoSort sorters; 18 more Specimen Processing work stations Open in new tab The elapsed time from our learning of MDS AutoLab (January 1995), which was the first manufactured TLA system we believed might meet our specific needs, until the “go live” date with this system was nearly 4 years. More than 2.5 years were devoted to evaluation, planning, and design. A contract with MDS AutoLab to build a system was not signed until November 1997. While the system was being constructed at the factory, a joint project team worked out the detailed specifications for the interfaces from ESP to APX and APX to ESP. Also during this time, a database was built in APX, consisting of the test codes and test names to be routed to each AutoSort and the templates of test codes for each lane of each sorter. More than 1700 different test codes were built for the automation system and reviewed for accuracy by multiple individuals. In addition, three ARUP employees volunteered to be the first principal users (Super Users) of the system and were trained extensively at the factory. The system was built, installed in our laboratory, and declared operational by AutoLab ∼8 months after we signed the contract. However, there was an additional delay of 4.5 months until implementation as we continued to make critical improvements to ESP and then tested these upgrades with the automation system. Seven versions of ESP Accessioning and numerous versions of other ESP modules were released between March and November 1998. During this time, most specimen processors used Pathnet for accessioning, whereas ESP was tested by a small group of experienced processors who could report on errors and functional needs that were yet required. Finally, in late October 1998, we believed that we had a version of ESP that could both support the automation system and adequately function for accessioning, and we began to train all specimen processors on ESP Accessioning. All employees who were to interact with the AutoLab system were trained by AutoLab personnel or by our Super Users. We then began using the automation system in production on November 17, 1998. 1 " Nonstandard abbreviations: TLA, total laboratory automation; STC, standardized tube carrier; AGV, automated guided vehicle; ESP, expert specimen processing; and LIS, laboratory information system. 2 " The corporate names used here were the names in use at the time of our evaluation in 1996. Coulter is now Beckman CoulterTM and Boehringer Mannheim Diagnostics is now Roche Diagnostics Corporation. We gratefully acknowledge the participation and support of literally hundreds of employees at ARUP Laboratories without whose involvement, hard work, and positive attitude a project of this magnitude could not have been undertaken. The success of this project speaks to that effort. We also acknowledge the major contributions of a large number of employees of MDS AutoLab in the early consulting and evaluation stages of this project, including the development of the work flow map: Mary Dunphy, Peter Payette, Sharon Cousineau, Susan Kewin, and Kathy Boroski. We also gratefully note the efforts of Devon Piirto, Rob Gordanier, Paul Dean, Hubert Thomas, Alex Stefou, and Alex Ciraco of MDS, and Murray Taylor of Automation Tooling Systems in the key stages of design, testing, and installation of the automated system and in support of the system after its installation. We also thank Tim Villnave of Workplace Ergonomics (Salt Lake City, UT) for review and study of the ESP workstation design, and cabinet maker Blaine Parry (Bluffdale, UT), who constructed all of the ESP workstations. References 1 Markin RS. Recent trends in clinical laboratory automation. Clin Lab Manage Rev 1998 ; 12 : 176 -180. PubMed 2 Sasaki M, Ogura K, Kataoka H, Imamura J. Automated clinical laboratory systems. Kost GJ eds. Handbook of clinical automation, robotics, and optimization 1996 : 442 -467 John Wiley & Sons New York. . 3 Kawai T. The status of laboratory automation in Japan [Abstract]. First Cherry Blossom Symposium on Clinical Laboratory Automation and Robotics 1998 : 37 Kochi Medical School Kochi, Japan. . 4 Bauer S, Teplitz C. Total laboratory automation: a view of the 21st century. Med Lab Observer 1995 ; 27 : 22 -25. 5 Hawker CD, Garr SB, Hamilton LT, Ashwood ER, Weiss RL. An automated system for transporting and sorting laboratory specimens: the ARUP plan. J Assoc Lab Autom 1998 ; 3 : 38 -40. 6 Simson E. A total laboratory automation system with fully integrated laboratory information system [Abstract]. Lab Automation ’98 1998 : 69 Association for Laboratory Automation Charlottesville, VA. . 7 Bauer S, Teplitz C. Total laboratory automation: system design. Med Lab Observer 1995 ; 27 : 44 -50. 8 Middleton S, Mountain P. Process control and on-line optimization. Kost GJ eds. Handbook of clinical automation, robotics, and optimization 1996 : 515 -540 John Wiley & Sons New York. . 9 Hawker CD, Roberts WL, Garr SB, Hamilton LT, Penrose JR, Ashwood ER, et al. Automated transport and sorting system in a large reference laboratory: part 2. Implementation of the system and performance measures over three years. Clin Chem 2002 ; 48 : 1761 -1767. Crossref Search ADS PubMed © 2002 The American Association for Clinical Chemistry This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model)
TI - Automated Transport and Sorting System in a Large Reference Laboratory: Part 1. Evaluation of Needs and Alternatives and Development of a Plan
JF - Clinical Chemistry
DO - 10.1093/clinchem/48.10.1751
DA - 2002-10-01
UR - https://www.deepdyve.com/lp/oxford-university-press/automated-transport-and-sorting-system-in-a-large-reference-laboratory-X6nx1jufPP
SP - 1751
VL - 48
IS - 10
DP - DeepDyve
ER -