Operations Research “ from blood withdrawal to first results” in the routine 24/7 clinical chemistry laboratory
Prof. dr. Richard J. Boucherie, CHOIR, University of Twente
Prof. dr. Yolanda B. de Rijke, Clinical Chemistry, Erasmus MC, Rotterdam
Prof. dr. Robert de Jonge, Clinical Chemistry, VUmc en AMC, Amsterdam
Drs. K. Vroonhof, Consultancy, Roche Diagnostics
Dr. Hans Schotman, Clinical Chemistry, VUmc, Amsterdam
Dr. Johan Fischer, Clinical Chemistry, AMC
Dr. Christiaan Raamakers, Clinical Chemistry, Erasmus MC
Dr.ir. N.M. van de Vrugt, University of Twente
Dr. Fenna Heyning, Director STZ hospitals (advisor)
To be determined
The routine 24/7 clinical chemistry laboratory has undergone a trend towards automation in recent years. Aim of the automation was to reduce human error (e.g., switching of samples during manual pipetting), to improve turn-around-time, to have predictable performance, to increase capacity, and to reduce costs. Initially, the analytical phase was automated with the introduction of “random access” analyzers, who were capable of analyzing selected tests in short time. Thereafter, the pre-analytical phase was automated (e.g., CLAS1 and Modular) and connected to the analytical phase. Functionality include automated registration, decapping, centrifugation, aliquoting, recapping and sorting. The last generation “total laboratory automation” (TLA) systems has also automated and connected the post-analytical phase (Erasmus MC). After pre-analysis and analysis, samples are automatically placed in a refrigirated archive, where they can be automatically retrieved to perform re-run testing when ordered. Together with these TLA solutions, also the software (middle ware systems, e.g. CITM) controlling the TLA system has become more and more important. The software controls the flow and logistics of the tubes in the TLA system, has track&trace functionality, and integrates confirmation and medical validation of test and quality control results. The newest software is also capable of Business Intelligence tooling.
Traditionally, the 24/7 laboratory was organized in functional units such as chemistry and hematology, et cetera. The last years, greater focus on process-oriented work flows has emerged both in the clinic (e.g. the integrated practise unit in value-based health care) and also laboratories. Process-oriented workflows better match customer (patient and doctor) demands and ask for critical process indicators (KPI’s) as well as outcome indicators to monitor the performance of the process. Current middle-ware solutions of TLA concepts are not equipped with tools for real-time monitoring of KPI’s and outcomes. Most Dutch laboratories are ISO 15189 accredited in which prospective risk assessment and monitoring of related KPI’s is important. Hardly any scientific literature is available on which critical KPI’s are important to monitor in TLA systems used in clinical chemistry. Furthermore, different TLA concepts are available on the market: from stand-alone machines, to work eilands, to completely open and integrated systems where all analyzers of different suppliers are connected to a track system. The TLA configuration differs per installation site, mostly based on best practises, customer demands and “gut-feeling”. However, the choice for a specific TLA set-up matching the laboratory case-mix and hospital demands has hardly any scientific foundation.
Aims of this study
In this operations research investigation, we will examine the process from blood withdrawal to first results in the routine 24/4 clinical chemistry laboratory. With the acquired knowledge, we aim to improve the routine 24/7 clinical chemistry process. This investigation has the following aims:
a)To investigate the current characteristics of the process from blood withdrawal to first results in the LIS in the clinical chemistry laboratory. To this aim we will compare routine chemistry (Roche, Cobas 8000/8100/Modular MPA/P512/P612) and hematology (Sysmex XN-9000) processes between the three participating academical centers (Erasmus MC – AMC – VUmc). These academic sites all have the same equipment but different configurations. By comparing the configurations at the different sites and taking into account the case-mix we would like to learn the basic performance characteristics of different configurations.
b)With the acquired information (a), we will experiment with different configurations such as a completely integrated hematology-chemistry workflow using mathematical modelling. This can be extended to routine hemostasis.
c)To investigate difference in prioritized (CITO) vs non-prioritized (non-CITO) sample logistics by the three different configurations. What is the effect on waiting time for non-prioritized samples? Should CITO’s be dealt with seperately or is one single process for prioritized and non-prioritized samples faster. How to deal with stat samples?
d)To investigate wich KPI’s have the greatest impact on the performance/outcome of the process or show the greatest variability and hence, should be monitored. The aim is to assemble a critical set of KPI’s that should be monitored by the middle-ware (dashbord; see ISO 15189).
e)To investigate the front-service where blood sampels are withdrawn and pre-analyzed. All three participating centers are consolidating front-service activities for clinical chemistry, pathology, medical immunology and infectious disease laboratories. Which workflow generates the best and fastest results?
•We will start by comparing the processes (24/7 chemistry and hematology laboratory) of the three participating academical sites Erasmus MC, AMC and VUmc. These sites all have the same suppliers (Roche, Sysmex) but different configurations and workflows. Depending on the progress of the project, we can extend the analysis to include a secondary care hospital with a different case-mix and configuration (OLVG or LabNoord).
•The PhD student will be appointed at the University of Twente and must have a background in Operations Research / Operations Management or Business Administration. The PhD student will be imbedded in the knowledge center CHOIR (Center for Healthcare Operations Improvement & Research, see www.utwente.nl/choir) of the University of Twente. For this project, the PhD will also be appointed in the three participating centers ( 0-aanstellingen) to arrange access to all required facilities (library, Email, et cetera) and to comply with local regulations, safety rules and research ethics.
•The partly present and partly to be developed Operations Research methods fall within the area Stochastic Optimisation, with aspects falling within Markov decision theory and integer linear programming. For reference projects using this methodology, see the project of Imke Gerritsma regarding the organisation of the process of next generation sequencing in the University Hospital Groningen (UMCG; zie http://essay.utwente.nl/67834/1/Gerritsma_MA_EEMCS.pdf) and Gréanne Leeftink regarding the organization of efficient pathology diagnostics in the University Hospital Utrecht (UMCU, see A.G. Leeftink, R.J. Boucherie, E.W. Hans, M.A.M. Verdaasdonk, I.M.H. Vliegen. P.J. Van Diest. Predicting turnaround time reductions of the diagnostic track in the histopathology laboratory using mathematical modelling, Journal of Clinical Pathology and A.G. Leeftink, R.J. Boucherie, E.W. Hans, M.A.M. Verdaasdonk, I.M.H. Vliegen. P.J. Van Diest. Batch scheduling in the histopathology laboratory. to appear Flexible Services and Manufacturing).
Furthermore, data will be analyzed using statistical methods.
•The PhD student will be supervised by three promotores from the participating centers (U Twente: prof. Boucherie, VUmc/AMC: prof. De Jonge, Erasmus MC: prof. De Rijke) and has two co-promotores from the University Hospitals. The promoters from the University Hospitals are also Head of the entire department of Clinical Chemistry.
•The PhD student will start within VUmc and thereafter in AMC and Erasmus MC. This order is based on increasing complexity and volume to allow the PhD student a complete view of the 24/7 clinical chemistry laboratory. At each hospital location, the PhD will fall under the direct responsibility of the Department Head and promotor.
•We will start with a T0 measurement of the chemistry and hematology TLA systems (order of logistics, time registration/time stamps/configuration principles such as CITO samples and batches) in all three centers. On this basis, we will identify rules/pinciples of the configuration of processes. How do choices in the configuration influence the process? This information can also be used to simulate new work processe/configurations (e.g., connected chemistry and hematology; connected P612; best way to reduce the TAT time, et cetera).
•The results can be partly retrieved from the middle ware or LIS systems (e.g., time stamps). Roche Diagnostics will facilitate with extraction of csv files from the analytical systems. Other processes should be investigated using more manual techniques (e.g., blood withdrawal to P612).
The results of the investigations will be published in peer-reviewed literature (operations research, clinical chemistry).