Optimization on reagent-loading manner for modular clinical chemistry analyzer series: simulations and verifications

Laboratory setup

In this study, we used the modular analyzer series, which includes p610 aliquoter and five analyzers. These analytical systems were used to test the reagent-loading mode effects on analytical units in MA. The p610 opens lids of the sample tubes, reads the racks, sample identifications and sorts tubes within the rack [10]. We chose one modular analyzer series (Roche Diagnostic, cobas 8000 c701 X 2 c502 X 1 e602 X 1) for better studying the reagent-loading manner effects on tTAT. One cobas 8000 analyzer contains two identical c701 modules, which comprise four disks and 140 reagent bins in total.

Data analysis and simulation verifications

Our project was divided into three processes. Firstly, we analyzed current reagent setup (Pattern 1) of the analyzer series, especially for the reagent-loading parameters. Secondly, simulations for reagent-loading optimization, aiming to improve the efficiency and cost reduction, proceed with cobas 8000 workflow simulator. cobas 8000 workflow simulator was used to retrospective and prospective analysis, simulation for one configuration of the cobas 8000 modular analyzer and gives answers to TAT. cobas 8000 workflow simulator includes four functionalities: overview interface, simulation interface, view output interface and setup interface. The analyzer data was the actual laboratory data and collected as CSV format from data manager, which contain sample ID, tests, sample ID reading time and sample unloading time. Then, these data was input into the simulator for configuration defining (overview interface). Modules type choices and mapping, data mapping, tests matching and distribution, mapping of analytes were operated in simulation interface. Analytical tests in one modular are corresponding to reagents’ types. This step, containing tests matching and distribution, was important in this work. Because it involves reagent situation in the modules’ reagent disks and eventually, decides reagent-loading manner. It is worth noting that user-defined reagents were replaced by similar vendor-supplied reagents in the simulation process. Once these steps above have done, cobas 8000 workflow simulator creates a named project. Then, simulation results for one project were calculated in the view output interface. It mainly includes the tTAT-1 and tTAT-2, which could be used for better comparisons with different projects. Pattern 2 and Pattern 3 were theoretical reagent-loading design schemes. TATs in these two theoretical reagent-loading manners were simulated based on Pattern 1. Parameters and differences were derived by cobas 8000 workflow simulator. Finally, we chose Pattern 2 for the simulation verification and it works best until now.

Simulation foundation – the Tuesday effect

Patients in the First Hospital of Zhengzhou University have a rhythm (Figure 1), which we call the Tuesday effect and termed as specimen summit. People always choose to make an appointment with the doctor through the outpatient service on Monday. Then, the doctors’ advices were given to the patients and some of them would be hospitalized depending on the diseases and became inpatients. On the Tuesday’s morning, routine tests would be conducted aiming to evaluate the present patients’ conditions. These tests were clinical chemistry related assays and represented by liver and renal function assays, routine chemistry panel, blood glucose, myocardial enzymes and lipids profile listed in Table 1. Due to the Tuesday effect, 3,730 ± 100 clinical chemistry specimens would be tested and they had similar test profile on every Tuesday. Tuesday blood specimens account for 19.61 ± 0.44% of total specimens in one week. The analytical unit studied here was one of the five cobas 8000 analytical analyzer series and the serums were sorted equally by p610 aliquoter. It is impossible to use all the same samples in one Tuesday to repeated test or verified in another day for differentiating the reagent-loading modes. But, this rhythm in clinical serum samples supports that our study can get almost the same serum test conditions in different Tuesdays. Ultimately, we could verify the cobas 8000 workflow simulator result at the same condition to evaluate the reagent-loading manner.

Figure 1:

The tendency chart and the Tuesday effect of clinical chemistry serums in our clinical chemistry laboratory within 62 days. Our test tube number was between 300 and 800 (almost 5,000–15,000 clinical chemistry tests) within a week. Black triangles represent the serum numbers on Tuesday and black squares were the other day. The tTAT and reagent-loading pattern before optimization.

Pre-analytical related TAT includes centrifuging, charging, racking, transporting, serum tube distribution and test assignment [11]. Analytical related tTAT was effect by instruments, analytical methodologies, and reagent-related factors [12]. Here, we focused on analytical related tTAT, and studied the reagent-related factors effects on TAT. tTAT covered the average TAT between the first rack being sent into analyzer and the last test results in the rack being sent to the LIS. For detail, tTAT contained two statistical patterns, tTAT-1 and tTAT-2 (Figure 2). tTAT-1 represents the average time from the sample loading or registration to rack unloading. tTAT-2 covered the time points from rack unloading to last result sending in the rack.

Figure 2:

The workflow in our clinical chemistry laboratory and the flow dissections of tTAT.

Due to the analytical requests of the in- and out-patients and the Tuesday effect, we incessantly imported five modular analytical series during 2007 and 2012. We chose one series as research model, which contains one ISE 900 electrolytes module, two c701 and one c502 clinical chemistry modules, and one e602 immunechemistry module. Blood electrolytes, clinical chemistry and immunechemistry of requested tests account for 9, 90 and 1% (Supplemental Figure 1), respectively. Thus, electrolytes and clinical chemistry requested tests were the main factors for tTAT. Due to the rapid test for electrolytes, only the clinical chemistry modules represented by c701 units were our research focus.

The c701 unit is composed of two parts, disk A and disk B, each of which has 35 reagent bins. The c502 module contains 60 reagent bins. Our analyzer series has two chemistry units and 140 reagent packs in total. For one series, we always employ Pattern 1 which ensured that at least three reagent kits of one test could be used at the same time for two c701 (Supplemental Table 2). This mode remains effective in our lab, and ensures that every cobas 8000 achieves the shortest tTAT and comparably uniform species distribution. For this reagent-loading mode, average tTAT-1 and tTAT were listed in Table 2 (Pattern 1).

We took it for granted that Pattern 1 can not only reduce reagent loading time but also save the labor. In fact, Pattern 1 was laborious during use due to loading reagents on the basis of four disks and 140 reagent bins for almost 40 clinical chemistry tests. Considering this situation, we established a standard operation process, and ran from 8:00 a.m. to 5:30 p.m. We conducted all the tests from 8:00 a.m. to 2:30 p.m. and reluctantly conducted reagent loading, calibration and the instruments’ maintenance from 2:30 p.m. to 5:30 p.m. It was labor-consuming and time-costing to load reagents for 3 h in Pattern 1. Most importantly, the clinical process was interrupted by reagent loading. These time should have been conducted another more tests instead of leaving it until tomorrow. Top 10 clinical chemistry tests of Pattern 1 were listed (Supplemental Figure 2). The average on-line time of these reagents was 7.6 ± 4.39 days (Table 2), which is a problem for reagents with short on-board stability, such as ALP, which remains stable only for 7 days at 2–8 °C. The QC cost for Pattern 1 was 35.75 Chinese Yuan. Test item performance verification times for one modular analyzer series in our lab were conducted by reagent disks. The experimental verification of Pattern 1 needs at least three times for one testing item and this was very complicated in practice.

Two theoretical loading modes

Before implementing the reagent-loading project, we conducted theoretical simulations on Pattern 2 (Supplemental Table 3) and Pattern 3 (Supplemental Table 4) based on the practical Pattern 1 data. We aim to optimize our reagent-loading project to shorten the tTAT and expenses. Besides, the simulations made the best effort to reduce the reagent packs for every c701 test item, as well as the costs of performance verification and labor. All of these optimization designs were conducted as a prerequisite of at least maintaining the available efficiency of Pattern 1. Thus, TAT simulation results show that Pattern 2 would be the best choice for our lab (Table 2). Furthermore, other parameters in Table 2 reveal that Pattern 2 could be put into practice.

Pattern 2 verification

tTAT and tTAT-1 for Pattern 2 and Pattern 3 in Table 2 were theoretical basis calculated from the cobas 8000 simulator. The actual tTAT for Pattern 2 should be tested after optimization mode being confirmed. Before the transformation of reagent-loading manner from Pattern 1 to Pattern 2, we trained all of our technicians to make sure that all the reagents can be loaded in Pattern 2. Then we ran data in Pattern 2 under the same conditions as Pattern 1 according to the Tuesday effect. tTAT and tTAT-1 were 1,020 ± 202 and 1,502 ± 234 s, respectively. In practice, tTAT and tTAT-1 of Pattern 2 reduced 61 and 49% than that of Pattern 1. It means that clinical chemistry results can be sent more rapidly and improve patients’ satisfaction. Reagent on-line time in Pattern 1 decreased 43% in total. This ensures the robustness of the results effectively. Furthermore, QC cost and performance verification times save the expenditures for 43%. The lower labor elevated the laboratory efficiency and improved employees’ satisfaction. In conclusion, Pattern 2 gets better effects.