| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Articles |
1
Stichting Artsen Laboratorium, Noorderstraat 8, 3512 VX Utrecht, The Netherlands.
2
Diakonessenhuis, Bosboomstraat 1, 3582 KE Utrecht, The
Netherlands.
a Author for correspondence. Fax 31-30-2361172; e-mail S.A.L.Utrecht{at}inter.nl.net
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Methods: A workload protocol was developed and used to assess processing capacity and efficiency parameters of five immunoassay analyzers under standardized conditions in a real-life routine situation. We studied the ACS:Centaur® (analyzer A), ArchitectTM i2000 (analyzer B), Elecsys® 2010 tandem (analyzer C), Immulite® 2000 (analyzer D), and Vitros ECi (analyzer E) on the basis of a standardized workload protocol that reflected a routine laboratory situation. This workload encompassed reflex and STAT testing, dilutions, and in-run calibration of a new reagent lot number. The analyzers were compared for hands-on labor time, unattended time (UT), throughput, and differentiated relative productivity indexes [RPI(UT); number of reportable results/(processing time - sum of unattended time)]. The RPI data for analyzers linked to an automated (aut) sample-handling system [RPI(aut)] were also calculated.
Results: The evaluation produced a set of parameters for the productivity of the instruments. An overview of the most important parameters revealed the following: the throughput was 193, 123, 97, 109, and 46 tests/hour for instruments A, B, C, D and E, respectively; the RPI(10) was 425, 238, 161, 445, and 151 tests/operator-hour; the RPI(30) was 229, 136, 118, 264, and 86 tests/operator-hour; the RPI(10,aut) was 1701, 637, 235, 964, and 223 tests/operator-hour; and the RPI(30,aut) was 298, 150, 174, 400, and 114 tests/operator-hour.
Conclusions: The combination of a standardized workload protocol and determination of parameters for productivity and labor efficiency, especially the differentiated RPIs, made it possible to make an objective comparison of the organizational consequences of the use of these instruments. The described parameters allow for a scientifically based choice, given a certain workflow and a particular laboratory organization.
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
The particular requirements that new analyzers must meet in a given laboratory situation can be formulated after analysis of that laboratory. Good analytical performance is a prerequisite, but the throughput, efficiency, and labor requirements must meet the needs of the laboratory. In view of the diversity in laboratory organizations, these requirements can vary greatly, but little information has been published about processing capacity and efficiency of analyzers. With the trend toward enlargement of laboratory organizations and consolidation of workstations, studies are needed to evaluate workflow, labor requirements, and solutions for coping with workload and test diversity. In this study, productivity and efficiency of the analytical process of five different analyzers were evaluated by measuring a standard workload that encompasses reflex and STAT testing, dilutions, and in-run calibration of a new reagent lot number. This workload reflected the frequency and distribution of immunoassays as performed in many laboratories in The Netherlands. The efficiency and processing capacity were expressed with the following parameters: throughput (results/hour), the differentiated relative productivity index [RPI(UT); results/operator-hour], and hands-on labor time (minutes). The RPI, as defined by Brzezicki (5) and Girgensohn et al. (6), provides information only about a theoretical situation in which all time periods during which an operator does not actively work with the system can be used for other activities. Because in a routine laboratory situation very short time periods cannot be used, this RPI [RPI(0)] is not very meaningful. In the present study, the RPIs were differentiated by taking into account the variable lengths of unattended time that are useful for performing other tasks in the different laboratory organizations. In addition, the volume and weight of waste products were described.
The objectives of this study were to determine, under comparable conditions, in which ways the characteristics of the examined analyzers meet productivity and labor time requirements.
|
Materials and Methods |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
The fully automated random access analyzer ArchitectTM i2000 (software Ver. 1.00; Abbott) uses chemiluminescent immunoassay technology that incorporates a acridinium derivative tracer. This floor model uses paramagnetic microparticles as solid phase. After exposure to pretrigger and trigger reagent, the acridinium undergoes a decomposition reaction and the emitted light is amplified and processed. The Architect has 25 reagent pack positions.
The tandem rack version of the Elecsys® 2010 system (Roche-Boehringer Mannheim), comprising two rack-sampling Elecsys 2010 analyzers connected to a PC-based software package called Laboratory System Manager (LSM; Ver. 2.41) (7) was used to measure productivity. The protocol was run on both machines in parallel, and the assays were distributed over the systems: thyroid-stimulating hormone (TSH), free thyroxine (FrT4), prostate-specific antigen, and ferritin in system A; and TSH, FrT4, vitamin B12, folate, luteinizing hormone, follicle-stimulating hormone, estradiol, prolactin, progesterone, carcinoembryonic antigen, and human chorionic gonadotropin (HCG) in system B. The fully automated benchtop analyzer Elecsys 2010 (software Ver. 3.08; Roche-Boehringer Mannheim) incorporates an electrochemiluminescence detection cell. The streptavidin-coated paramagnetic beads are coupled to the ruthenium-labeled antigen-antibody complex. After the addition of tripropylamine, a voltage is applied, and the resulting luminescence is measured (8)(9). The Elecsys 2010 tandem has 30 reagent positions and 24 different assays, which are available simultaneously.
The assays of the floor model, fully automated continuous random access analyzer Immulite® 2000 (software Ver. 1.2; Diagnostic Products Corporation) are based on an alkaline phosphatase label and a chemiluminescent substrate and use a centrifugal wash method. The Immulite 2000 has 24 reagent positions (10)(11)(12).
The Vitros ECi (software Ver. 2.0; Ortho Clinical Diagnostics) is a fully automated random-access immunoassay system that utilizes an enhanced chemiluminescence technology. This floor model analyzer uses streptavidin-coated plastic wells as solid phase. Horseradish peroxidase is used as label, and a luminogenic substrate (luminol derivative and peracid salt) is used for signal detection. The Vitros ECi has a reagent capacity of 20 reagent packs (13).
samples
Anonymous patient sera were collected, aliquoted, and stored at
-20 °C. To exclude effects of rethawing, a set of 400 aliquots was
made available for each analyzer to perform the workload study. The
samples were stored in the same manner for all of the analyzers to
avoid the introduction of storage conditions as a variable. The samples
for each experiment were thawed and homogenized immediately before the
experiment commenced.
In the beginning and at the end of the workload, high, medium, and low controls (Lypocheck Immunoassay Plus Control Levels 1, 2, and 3) were measured for all 13 parameters to check proper performance of the system.
workload protocol
Introduction.
The workload protocol was developed by
comparison of the immunoassay workload of four medium-sized hospital
laboratories and a general practice laboratory for assays, number of
tests per tube, STAT tests, and reflex tests. No essential
differences in request pattern appeared to exist between these
different Dutch laboratories. Only the total number of requests varied.
Protocol.
The workload protocol consisted of ~700 requests,
400 tubes, and 39 quality-control measurements in 13 different assays:
FrT4 (15%) and TSH (45%) for thyroid function; folate (6%), vitamin
B12 (7%), and ferritin (6%) for anemia;
prostate-specific antigen (6%) and carcinoembryonic antigen (1%) as
tumor markers; and follicle-stimulating hormone (3%), luteinizing
hormone (3%), estradiol (2%), progesterone (2%), prolactin (2%),
and HCG (1%) for fertility. The percentages given are the
contributions to the total number of requests.
At the time of this study, some assays were not available and were replaced by other assays: vitamin B12 replaced folate in the Architect i2000; third-generation TSH replaced vitamin B12 and ovarian marker monoclonal antibody (CA 125) replaced folate in the Immulite 2000; TSH and vitamin B12 replaced carcinoembryonic antigen and prostate-specific antigen and anti-hepatitis B surface antigen antibodies replaced folate for the Vitros ECi.
Instrument productivity was additionally challenged by changing a TSH reagent lot number and calibration halfway through the work list, performing dilutions, doing STAT requests, and carrying out reflex tests:
The procedures for the reflex tests were as follows. For every TSH out of the reference range (0.2–3.5 mIU/L) a FrT4 reflex test was performed. In addition, to simulate the reflex tests that are usually requested via LIS for deviating hematology results, vitamin B12, ferritin, and folate requests were performed for every luteinizing hormone request.
To test the effect of dilutions, the workload contained two samples that needed dilution: a sample tube with a high (~200 mIU/L) TSH concentration and a sample tube with a high (~100 000 IU/L) HCG concentration.
To test the impact of STAT requests, STAT samples were presented 20 and 80 min after the workload was started, these being an estradiol sample and a high HCG sample that needed dilution, respectively.
To test the effect of a change of reagent lot number and calibration, before the workload procedure was started, the systems were supplied with full reagent packs, except for the TSH and ferritin reagent packs, which had enough reagent for 70 and 12 requests, respectively, forcing an in-run change of reagent and recalibration.
Protocol procedure.
Work lists were entered into the
instruments manually. However, because in routine situations work list
entry is accomplished by the LIS and the time required for this task
depends on the type of LIS, the time needed to enter work lists was not
included in the processing time used in the calculations of the
results. The sample tubes were bar-coded and placed on the work bench
in standard tube racks. Sample handling, therefore, included placing
samples in the system’s own racks and removing them after completion
of the analysis. All types of handling that the operator performed on
the systems during the analytical phase of the workload were timed and
recorded.
Definitions of productivity and efficiency parameters.
The
following parameters were measured or calculated (5)(6):
 |
 |
A differentiation of RPI was made by the different unattended time (UT) values given. The length of the unattended time depended on the laboratory organization.
|
Results |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
. The length of time from the time the analyzer was started to
the first results was 21 and 22 min for the ACS:Centaur and the Elecsys
2010 tandem, respectively. The length of time to the first result for
the Architect was 38 min. The Vitros ECi reported the first result
after 42 min, and the Immulite 2000 reported the first result after 45
min, which mostly reflected the longer incubation time that was
required.
|
The first STAT sample introduced during processing was analyzed for estradiol. The second STAT sample was analyzed for HCG and required further dilution to obtain the final result. The time required from the receipt of the STAT estradiol until the result was reported was 19 min for the ACS:Centaur, 34 min for the Architect i2000, 23 min for the Elecsys 2010, 73 min for the Immulite 2000, and 43 min for the Vitros ECi. The time from receipt of the STAT HCG sample until the final result was reported was 40 min for the ACS:Centaur, 60 min for the Elecsys 2010 and Vitros ECi, and 75 min for the Architect i2000 and Immulite 2000.
hands-on labor time
Total hands-on labor time and the distribution of the hands-on
labor required for performance of different labor tasks are represented
in Fig. 2
. Of the instruments compared, the ACS:Centaur needed the least
amount of hands-on labor time.
|
During our workload experiment, interruptions of the routine analytical process on the ACS:Centaur and Vitros ECi occurred.
The amount of labor required for performance of STAT tests and calibration procedures of the five systems was comparable, and reagent handling times were almost negligible on all five analyzers.
The ACS:Centaur, Architect i2000, and Immulite 2000 required considerably less hands-on operator time for the handling of disposables and waste than the Elecsys 2010 tandem and the Vitros ECi, both of which required similar lengths of time for this part of the process.
The lengths of time required for placing the samples in the analyzer racks and removing them after completion were virtually the same for all five analyzers. For the ACS:Centaur, Architect i2000, and Immulite 2000, ~60% of the total hands-on labor time was required for sample handling. For the Vitros ECi and Elecsys 2010 tandem, sample handling accounted for 30% (36 min) and 20% (23 min), respectively, of the total hands-on labor time.
Almost 50% (45 and 67 min, respectively) of the Elecsys 2010 and Vitros ECi hands-on labor time was used for selecting samples that required reflex tests and dilutions. The Architect i2000 and Immulite 2000 needed 14 and 10 min (20% and 14%), respectively, to perform the reflex tests and dilutions. On the ACS:Centaur, no operator time was required for these tasks.
throughput
The throughput data with the length of time needed for instrument
initialization included, the throughput data with the length of time
needed for instrument initialization excluded, and the throughput data
as specified by the suppliers are shown in Table 1
.
|
rpi
The productivity results are given in Fig. 3
, expressed as RPI values vs
length of unattended time. The RPI value for the maximum length of
unattended time is by definition equal to the throughput because above
this maximum the sum of unattended time is zero. The maximum
productivity of labor (maximum RPI) is obtained when all walk-away
periods theoretically can be used to perform other processes, or stated
differently, when the length of unattended time approaches zero.
|
waste
As a result of this workload protocol, both the ACS:Centaur and
the Vitros ECi produced 8 L of liquid and 2 kg of solid waste. The
Immulite 2000 produced 8 L of liquid waste and 1 kg of solid waste. The
Elecsys 2010 produced 5 L of liquid waste and 1.5 kg of solid waste.
The Architect i2000 produced 26.2 L of liquid waste,
substantially more liquid waste than the other systems, but it produced
only 420 g of solid waste, the smallest amount of solid
waste.
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
processing time and stat time
Comparison of the total processing times of the analyzers for the
workload studied revealed that the ACS:Centaur required considerably
less time to finish all requests than the other four analyzers.
Interruptions during test processing were documented for the five
analyzers and are shown in Fig. 1
. During our study, operational
processing on the ACS:Centaur was interrupted once because of a
temporary incubation ring problem. This occurred only once during a
total evaluation period of 5 days.
Because of instrument design, reloading of liquid and disposables was not possible on the two Elecsys 2010 analyzers and the Vitros ECi while they were running. Accordingly, the assay processes had to be interrupted to replenish supplies. To change reagent packs, the Elecsys 2010 and the Architect i2000 also needed to be in pause mode, and therefore, on one of the Elecsys analyzers and on the Architect, the assay processing had to be stopped to add a new lot number of TSH reagent and perform the calibration procedure.
The ACS:Centaur and Elecsys 2010 rack analyzers have special STAT entry ports, and these entry ports were used during the workload study. STAT samples loaded via this entrance needed no extra requests to be handled with priority. On all of the analyzers, the tubes containing STAT samples could also be loaded in the usual way, but they needed to be requested as STAT. It took more time between receipt and sampling of the STAT samples on the Elecsys 2010 and the Architect (5–10 min for the Elecsys 2010 and Architect i2000 vs 1–3 min for the ACS:Centaur, Immulite 2000, and Vitros ECi). This can be explained by the fact that the Elecsys 2010 and the Architect i2000 had to complete all of the requests for the rack previous to the rack containing the STAT sample before they could sample the STAT request. The Elecsys 2010 has a feature for processing STAT HCG, with cycle times of 9 min instead of 18 min. During this study, this feature was not used. The Immulite 2000 is able to perform Rapid TSH and third-generation TSH, with incubation times of 30 and 60 min, respectively. During the evaluation, Rapid TSH was used.
hands-on labor time
Almost one-half of the hands-on labor time recorded for
troubleshooting on the ACS:Centaur was attributable to an incubation
ring malfunction, which occurred only once during the evaluation
period. By switching off-line only that part of the ACS:Centaur that
experiences the problem, other processing steps can continue, so that
troubleshooting rarely leads to lost results, minimally affects
throughput, and is easy to perform. The most frequent alarms on the
Elecsys 2010 during the evaluation period were related to problems with
the gripper, which performs transport of both sample tips and cups. The
availability of technical alarms or warnings in the software of this
system is minimal; for example, no warnings for short-term exhaustion
of reagents or water are displayed. During the study, a minimal amount
of time was required for problems, and the few error messages were
related to the transport of assay cups. Problems with the LSM occurred
during the latter half of the workload. Because of this problem, the
LSM was off-line at the end of the workload. This problem led to fewer
reportable results being achieved on the Elecsys 2010 tandem version in
comparison with the ACS:Centaur and Vitros ECi analyzers. On the
Architect i2000 and Immulite 2000, no time was spent on
troubleshooting.
The Vitros ECi displays four types of alarms: attention, action, malfunction, and shutdown. In the case of the first two alarm types, sample processing is not interrupted. The malfunction and shutdown alarms require operator interaction. On the Vitros ECi, the most frequently occurring problems were scheduling conflicts. In addition, difficulties in closing and opening of reagent packs as well as technical problems with the incubation ring occurred.
On the ACS:Centaur, Architect i2000, and Immulite 2000, liquid and disposable waste are removable during the run. To empty the liquid and disposable waste bins, the assay processes must be stopped on both the Elecsys 2010 and the Vitros ECi.
Especially for the ACS:Centaur, Architect i2000, and Immulite 2000, where sample handling took 60% of the total hands-on labor time, but also for the Elecsys 2010 and Vitros ECi (20% and 30%, respectively, of total hands-on labor time needed for sample handling), automation of sample handling by linking to automated sample-handling systems will yield large labor profits.
The reagent packs on all five of the analyzers are ready for use immediately, leading to only minimal time requirements for reagent handling. The ACS:Centaur has 30 reagent positions and continuous accessibility to enable reagent replacement. The Architect i2000 has 25 reagent pack positions, and reagent packs cannot be changed during the run. The Elecsys 2010 tandem has 30 reagent positions, and 24 different assays are available simultaneously, but reagent replacement is not possible during sample processing. The Immulite 2000 has 24 reagent positions and continuous accessibility for changing of reagent packs. The Vitros ECi has a reagent capacity of 20 reagent packs, and reagent replacement and calibration is possible during sample processing.
Because of the unattended autorepeat/autodilute instrument features, no hands-on labor time is required to perform reflex tests and dilutions on the ACS:Centaur. The racks of samples needing reflex tests or dilutions remain in the in process queue until the results of all requests have been determined. The Elecsys 2010 software is not able to perform reflex testing. This, therefore, can only be arranged through the LIS. On the Architect i2000, Immulite 2000, and Vitros ECi, adjusting for reflex testing is possible. However, the racks of samples for reflex tests and samples requiring dilutions still must be located, selected, and loaded for the second time on the Architect i2000, Elecsys 2010, Immulite 2000, and Vitros ECi.
throughput
There are two options available for sample processing on the
ACS:Centaur: the samples can be processed in sequence of presentation
of the tubes, or the assays can be processed in a calculated sequence
for the provision of optimal throughput. During measurement of the
workload, the ACS:Centaur was set to provide optimal throughput, while
holding samples for repeat or reflex testing. The differences between
the throughput data calculated including initialization time and the
maximum throughput data specified by the suppliers were 12%, 39%,
45%, 46%, and 49% for the ACS:Centaur, Architect i2000,
Elecsys 2010 tandem rack version, Immulite 2000, and Vitros ECi,
respectively. Without taking into account start-up time, the throughput
data on the ACS:Centaur, Architect i2000, Elecsys 2010
tandem rack version, Immulite 2000, and Vitros ECi differed 1%, 10%,
19%, 27%, and 32%, respectively, from the maximum throughput data
supplied by the manufacturers. The throughput of the Architect
i2000 and Elecsys 2010 tandem was more affected by
initialization time than the other systems. Differences in the
effective throughput data measured with this workload, compared with
the maximum throughput data supplied by the manufacturers, can be
explained by the need to perform dilutions, STAT requests, problem
solving, reflex testing, and changing and calibration of a new reagent
lot number, all of which decrease the maximum achievable throughput.
The necessary changing of reagent packs during a run in routine daily
practice can influence the efficiency of the analyzer greatly because
the effect of this intervention differs between instruments. This will
especially be the case as workstations become more consolidated; for
example, when serological and therapeutic drug monitoring parameters
can also be determined on this type of analyzer, and the number of
reagent positions on the analyzer is limited.
Girgensohn et al. (6) found on the ACS:Centaur an effective throughput of 188 results per hour by measuring a comparable workload. The mean output rate measured by Kunst et al. (14) on a single Elecsys 2010 analyzer was 70 results per hour. This is exactly one-half the throughput we measured on the two Elecsys 2010 analyzers.
rpi
In Fig. 3
, the labor productivity for the different immunoassay
analyzers is expressed as RPI vs length of unattended time. Both the
ACS:Centaur and Immulite 2000 appear to be labor-efficient analyzers,
producing RPI values that were substantially higher than those of the
other three analyzers over the whole range of unattended times. In
laboratories where, theoretically, the minimum time period that can be
used for other processes is 5 min [RPI(5)], the
ACS:Centaur demonstrates the highest production per operator, whereas
in a laboratory where only periods 
30 min are usable
[RPI(30)], the Immulite 2000 has a slightly
higher labor productivity compared with the ACS:Centaur. To simulate
the effect of linking the systems to an automated sample-handling
system on labor efficiency, RPI(aut) values were
calculated by adding the periods of hands-on labor time used for sample
handling to the unattended time (Table 2
). This produced obviously higher RPI(aut)
values for the ACS:Centaur, Architect i2000, and Immulite
2000 for unattended time lengths <25 min. The RPI values for the
Elecsys 2010 tandem and the Vitros ECi were also positively affected
but to a lesser extent.
|
useful productivity and efficiency parameters for planning
instrumentation in a laboratory organization
To make it possible to organize a laboratory in such a way that
one operator can be involved in different analytical processes, as is
seen mostly in small laboratories, there is a need for instruments that
can run unattended for long time periods. For these
laboratories, long unattended times are of more importance than high
throughput. In a laboratory organization where an operator is involved
in only one analytical process, it is more likely that operators can
use short unattended times productively, for example, for logistic
activities related to the process. This is found mostly in
large laboratories with large workloads. For these laboratory
organizations, high throughput is important, and the high RPI values
for short unattended times [RPI(5),
RPI(10)] are more applicable than for small
laboratories.
In large laboratories with automated sample-handling systems, operators are working as supervisors. Under these circumstances, high throughput as well as long unattended times are necessary.
In conclusion, the parameters measured, especially the differentiated RPIs measured by our application of a standardized workload protocol to evaluate and compare processing capacity and efficiency of immunoassay analyzers, make it possible to take a scientific approach toward instrument planning in a given laboratory situation.
|
Acknowledgments |
|---|
|
Footnotes |
|---|
|
References |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
The following articles in journals at HighWire Press have cited this article:
|
|
 |
|
  F. A. Quinn, M. Mittino, F. Rode, H. A. Hendriks, and W. M. Verweij Capacity and Efficiency Testing of New Immunoassay Analyzers Drs. Hendriks and Verweij respond: Clin. Chem., July 1, 2000; 46(7): 1017 - 1019. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
,
ACS:Centaur;
,
Architect; 
, Elecsys 2010 tandem;
, Immulite
2000; 
, Vitros ECi.