Journal of Medical Systems (2018) 42: 28

https://doi.org/10.1007/s10916-017-0878-1

SYSTEMS-LEVEL QUALITY IMPROVEMENT

Key Performance Indicators to Measure Improvement

After Implementation of Total Laboratory Automation Abbott
Accelerator a3600
Marijana Miler 1 & Nora Nikolac Gabaj 1 & Lora Dukic 2 & Ana-Maria Simundic 2

Received: 7 June 2017 / Accepted: 13 December 2017 / Published online: 27 December 2017
# Springer Science+Business Media, LLC, part of Springer Nature 2017

Abstract
The aim of the study was to estimate improvement of work efficiency in the laboratory after implementation of total laboratory
automation (TLA) by Abbott Accelerator a3600 in the laboratory with measuring different key performance indicators (KPIs)
before and after TLA implementation. The objective was also to recommend steps for defining KPIs in other laboratories. For
evaluation of improvement 10 organizational and/or technical KPIs were defined for all phases of laboratory work and measured
before (November 2013) and after (from 2015 to 2017) TLA implementation. Out of 10 defined KPIs, 9 were successfully
measured and significantly improved. Waiting time for registration of samples in the LIS was significantly reduced from 16 (9–
28) to 9 (6–16) minutes after TLA (P < 0.001). After TLA all tests were performed at core biochemistry analyzers which
significantly reduced walking distance for sample management (for more than 800 m per worker) and number of tube touches
(for almost 50%). Analyzers downtime and engagement time for analyzers maintenance was reduced for 50 h and 28 h per
month, respectively. TLA eliminated manual dilution of samples with extreme results with sigma values increment from 3.4 to >6
after TLA. Although median turnaround time TAT for potassium and troponin was higher (for approximately 20 min), number of
outliers with TAT >60 min expressed as sigma values were satisfying (>3). Implementation of the TLA improved the most of the
processes in our laboratory with 9 out of 10 properly defined and measured KPIs. With proper planning and defining of KPIs,
every laboratory could measure changes in daily workflow.

Keywords Key performance indicators (KPI) . Productivity . Total laboratory automation (TLA) . Quality

Introduction inevitable component of every modern laboratory [1]. TLA

Total laboratory automation (TLA) consolidates preanalytical
and postanalytical procedures in clinical laboratory and also
includes analytical systems for complete sample processing.
Although various TLA solutions are present in laboratories for
almost 30 years, in recent 10 years TLA is recognized as an
This article is part of the Topical Collection on Systems-Level Quality
Improvement
* Marijana Miler
marijana.miler@gmail.com
1
Department of Clinical Chemistry, Sestre Milosrdnice University
Hospital Center, Zagreb, Croatia
2
Department of Medical Laboratory Diagnostics, University Hospital
BSveti Duh^, Zagreb, Croatia
provides standardization of the sample management, increases
the workload efficiency with the optimization of laboratory
processes, and the most importantly, reduces errors in all
phases of laboratory work (preanalytical, analytical and
postanalytical) [2].
Key performance indicators (KPIs) are objective measures
for assessment of system efficiency and monitoring changes in
routine laboratory work [3, 4]. KPIs should be specific, mea-
surable, achievable, realistic and timely (SMART) [5]. KPIs are
usually defined as indicators of technical or organizational
changes (or combination of both). Purely technical KPIs mea-
sure, for example, performances of the new analytical system
while organizational KPIs include e.g. changes of the workflow
procedures. KPIs should be defined before any changes are
made in everyday processes and measured before and after
implementation of the changes. KPIs are needed in order to
objectively assess efficiency of the change processes [4].
28 Page 2 of 12
Before the TLA implementation, in the core laboratory
for the medical biochemistry and analytical toxicology
most of the sample management processes were manually
performed, e.g. assessment of the sample quality,
aliquoting and sample distribution. Besides that, laborato-
ry personnel had to walk a lot between rooms in the lab-
oratory to distribute samples on different analyzers.
Manual steps in the procedures were not standardized,
were subjective, time consuming and prone to errors [6].
In order to standardize processes, optimize workflow, in-
crease the productivity of personnel and to reduce the
possibility for errors, Abbott Accelerator a3600 (Abbott,
Abbott Park, IL, USA) TLA with complete solution for
preanalytical, analytical and postanalytical sample han-
dling was implemented.
The main goal of this study was to evaluate accom-
plishment of the TLA implementation by measuring
changes of 10 preanalytical, analytical and postanalytical
KPIs, before and after reorganization of the laboratory
work. Additional objective was to provide a flowchart of
steps required for evaluating implementation of TLA.
Materials and methods
Study design and setting
This study was conducted in the Department of Clinical
Chemistry (DCC) of Sestre Milosrdnice University
Hospital Center, accredited according ISO 15189 since
2007. In the laboratory, more than 350 tests are performed
with daily workload of approximately 1500 samples. DCC
had emergency and routine sample reception, with different
sample management. Emergency samples were admitted,
then centrifuged on the stand-alone centrifuges (Rotofix
32A, Hettich, Tuttlingen, Germany), registered to labora-
tory information system (LIS) and put on the core bio-
chemistry analyzers (Beckman Coulter AU2700 or
AU680). If necessary, after centrifugation samples were
manually aliquoted for other analyzers: Cobas e411
(Roche Diagnostics, Indianapolis, USA), Axsym (Abbott
Laboratories, Abbott Park, Illinois, USA) and Vitros 250
(Ortho Clinical Diagnostics, Buckinghamshire, UK). In
December 2014, total laboratory automation (TLA) by
Abbott Accelerator a3600 (Abbott, Abbott Park, IL,
USA) was introduced in the laboratory routine work.
TLA included automated track system with input-output
module, two integrated centrifuges, decapper, two
Architect c8000, one Architect i2000, Aliquoter and Sealer.
In order to estimate and compare work efficiency before
and after implementation of TLA, 10 key performance in-
dicators (KPI) were defined prior the introduction of
J Med Syst (2018) 42: 28
automation. All KPIs were measured in November 2013
and afterwards in different periods from 2015 to 2017.
Key performance indicators
1. Workspace area (WSA)
Workspace area was defined as a space directly associ-
ated with preanalytical, analytical and postanalytical man-
agement of the samples in the core biochemistry laborato-
ry. Width and length of workspace area were measured
with the laser rangefinder (Bosch DLE 40, Bosch, UK)
and area of each room was expressed in square meters.
Area where samples were processed, with all analytical
systems, centrifuges and space for sample quality assess-
ment was designated as (pre)analytical area, while area for
lab reports releasing was designated as postanalytical area
(computer rooms). Considering constant increase of labo-
ratory requests, our goal was to be able to process as many
samples by occupying lower amount of space. Remaining
(saved) space can be allocated for laboratory staff (reports
releasing, teaching, communication with clinicians).
2. LIS registration waiting time
Before TLA, processes of blood sample registration to the
LIS were different for emergency and routine samples.
Emergency samples were registered into LIS directly from
electronic requests from hospital information system (HIS),
immediately after arrival in the laboratory. On the other hand,
routine sample tests were requested on paper and registered in
batch, after registration of all received emergency samples.
Together with TLA, HIS was introduced in the entire hospital
and all tests requests were made electronically.
Measurement of waiting time was done during two
working days before (11th and 12th December 2013) and
after (23rd and 24th August 2016) TLA implementation.
For all received samples, time of arrival was recorded.
Afterwards, time of registration to LIS was exported from
information system. Waiting time from arrival to the reg-
istration of routine samples to the LIS was calculated and
expressed in minutes. Results were presented as median
and interquartile range.
Our goal was to decrease LIS registration waiting time.
3. Tests panel distribution
Tests panel distribution represented the number of tests mea-
sured outside of core biochemistry analyzers. Core biochemistry
analyzers were defined as analyzers on which the most of the
clinical chemistry tests were measured: AU2700 and AU680
(Beckman Coulter, Brea, USA) before TLA and analyzers with-
in TLA after implementation (two Architect c8000 and one
J Med Syst (2018) 42: 28
i2000; Abbott Laboratories, Illinois, USA). Cobas e411 (Roche
Diagnostics, Indianapolis, USA), Axsym (Abbott Laboratories,
Abbott Park, Illinois, USA) and Vitros 250 (Ortho Clinical
Diagnostics, Buckinghamshire, UK) were available as analyzers
outside core analyzers. Since samples with tests for other ana-
lyzers needed to be manually aliquoted and distributed, those
tests were identified as having potential for errors.
In November 2013 and February 2017 (before and after
TLA, respectively) all biochemistry tests requests were
counted from LIS. Afterwards, proportions of tests per
analyzers outside the core biochemistry analyzers were
calculated. As a measure of high quality performance of
processes, sigma value >4 was set.
Our goal was to decrease proportion of tests measured out-
side of core biochemistry analyzers.
4. Walking distance for sample management
Minimum number of laboratory staff required for daily
management of biochemistry samples was assessed. For daily
Meters walked per test ¼ ðsum of steps walked by all included workers during the day  average
distance o f 20 stepsÞ=average number of 30 most common test requested per day
For comparison of daily distance walked per worker, aver-
age number of tests requests after TLA was used:
Daily distance walked per worker ðmÞ
¼ ðtotal number of walked steps in day=number of workersÞ
average number of tests after TLA
average distance of 20 steps
Our goal was to decrease number of steps required for
sample management in order to reduce unnecessary waste of
energy and increase productivity of laboratory staff.
5. Tube touch moment (TTM)
Tube touch moment (TTM) was defined as an aver-
age number of touches necessary to process the sample
(tube) from the sample registration to LIS to the sample
archiving. TTM was counted for three most common
combinations of routine laboratory tests: samples with
tests for the core biochemistry analyzers only (A), sam-
ples with tests for the core biochemistry analyzers and
aliquots for other analyzers (B) and samples that only
needed an aliquots for other analyzers (C) (Table 1).
Before TLA, preanalytical system aliquoted only rou-
tine samples, while management (aliquoting and
Page 3 of 12 28
management of samples, several stages were recognized: ad-
mission of samples, registration to LIS, distribution to
preanalytical and analytical systems and to working stations
distant from the core biochemistry analyzers.
Number of steps for management of emergency and
routine biochemistry samples was measured by pedome-
t e r ( O m r o n p e d om e t e r H J - 7 2 0 I T C , H o o f d d o r p ,
The Netherlands) fixed at the belt of personnel during
the weekly working hours (7:30–15:30). Steps were
turned into meters. Firstly, 7 chosen representative staff
members crossed 20 steps. Afterwards, distance of those
20 steps were measured with laser rangefinder (Bosch
DLE 40, Bosch, UK). All measured distances were sum-
marized and divided with 7 to obtain average distance of
20 steps. Walking distance was measured in two periods:
in December 2013 (before TLA) and June 2016 (after
TLA implementation). Distance walked per test was cal-
culated for 30 the most frequently requested tests in LIS
according to the equation:
distribution) of emergency samples was done manually.
After TLA implementation all samples (routine and
emergency) were managed automatically, on TLA.
Proportions of samples with different combinations
of tests (A, B or C) were exported from LIS for two
days in November 2013 and February 2017. Proportions
of A, B and C combinations of tests were multiplied
with number of tube touches according to the following
equation:
TTM ¼ ðA number of tube touches  A proportionÞ
þðB number of tube touches  B proportionÞ
þðCnumber of tube touches  C proportionÞ
A samples with tests only for the core biochemistry
analyzers
B samples with tests for the core biochemistry analyzers
and aliquots for other analyzers
C samples that only needed aliquots for other analyzers
(C)
TTM tube touch moment
Our goal was to reduce number of tube touches.
6. Manipulation tube time (MTT)
28 Page 4 of 12 J Med Syst (2018) 42: 28

Table 1 Touches of tubes

necessary to process the sample Number Tube touch moment (TTM) Combination of laboratory tests
from registration to LIS to the
sample archiving before the TLA 1. Registration to LIS A, B, C
implementation 2. Barcode print A, B, C
3. Labeling primary tube A, B, C
4. Labeling aliquot tube B, C
5. Transfer to centrifuge A, B, C
6. Balancing centrifuge A, B, C
7. Putting on centrifuge A, B, C
8. Removing from centrifuge A, B, C
9. Decapping A, B, C
10. Aliquoting B, C
11. Putting on preanalytical system* A, B, C
12. Removing from preanalytical system* A, B, C
13. Assessment of interferences A, B, C
14. Transfer to other analyzers B, C
15. Putting on other analyzers B, C
16. Removing from other analyzers B, C
17. Putting on core analyzer A, B
18. Removing from core analyzer A, B
19. Putting on storage racks A, B, C
20. Archiving storage racks A, B, C

Legend: *only for routine laboratory samples, A – samples with tests for the core biochemistry analyzers only, B –
samples with tests for the core biochemistry analyzers and aliquots for other analyzers, C – samples that only
needed an aliquots for other analyzers

Manipulation tube time (MTT) per sample was defined
as duration (in seconds) of every stage counted as tube
touch moment (TTM). Duration for all procedures was
measured with stop-watch. To reduce variability of mea-
surement (due to different conditions) duration of every
stage was measured 20 times and at the end, average
duration was calculated. Three different MTTs were cal-
culated in emergency and routine biochemistry laboratory
according to the following equations:
1. MTT per sample = tube touch moment (TTM) x dura-
tion of all procedures x proportion of samples with the
most common combinations (A, B, C)
2. MTT per 500 samples (average daily number of sam-
ples) = MTT per sample × 500
3. MTT per day (in hours) = MTT per 500 samples / 3600
Our goal was to reduce duration of all tube touch moments.
7. Analyzers downtime (DT) per month
Analyzers downtime (DT) was defined as time while ana-
lyzers were out of function due to the maintenance procedures.
DTwas measured with stop-watch from the moment of analyzers
set in off-line (or shut down) until they were in the fully function-
ing mode (on-line). Analyzers included in the measurement
before TLA were: core biochemistry analyzers (AU2700/
AU680 (Beckman Coulter, Brea, USA), Cobas e411 (Roche
Diagnostics, Indianapolis, USA), Vitros 250 (Ortho Clinical
Diagnostics, Buckinghamshire, UK), preanalytical system
Beckman OLA 2500 (Beckman Coulter, Brea, USA), Axsym
(Abbott Laboratories, Abbott Park, Illinois, USA) and centri-
fuges. After TLA, measurement was done for 3 Architect ana-
lyzers: two c8000 and i2000, as well as for preanalytical system
Accelerator 3600 (Abbott Laboratories, Abbott Park, Illinois,
USA). DT was calculated for one month period by multiplying
number of maintenance operations (for daily, weekly and month-
ly maintenance, reagents filling, calibration and quality controls)
with time in minutes.
Our goal was to reduce total downtime of analyzers in one
month period.
8. Engagement time (ET) per month
Engagement time (ET) was defined as time needed for
active engagement of personnel for all maintenance proce-
dures with analyzers (filling with reagents, calibration, quality
control, daily, weekly and monthly maintenance). ET was
measured for analyzers from 7th KPI (DT) using stop-watch
in one month period in 2013 and 2017 and expressed in hours.
Our goal was to reduce the total time of personnel spent for
maintenance procedures.
J Med Syst (2018) 42: 28
9. Measurement ranges (MR) and dilution factors (DF)
Measurement ranges (MR) and proportion of samples with
values outside of measurement range were recorded during
one year period for enzymes: amylase (AMY) in serum and
urine, alanine aminotransferase (ALT), aspartate aminotrans-
ferase (AST), creatine kinase (CK) and lactate dehydrogenase
(LD) in serum at core biochemistry analyzers used before
(AU2700 and AU640) and after (Architect c8000) TLA im-
plementation. All analyzers used in 2013 and 2015 had auto-
matic dilution protocols, with different type of sample man-
agement. On AU2700/AU680, only one automatic dilution
for samples with enzyme activities above upper measurement
range was available. Samples had to be manually handled and
put on the analyzer for rerun in dilution. Also, automated
secondary dilution protocol was not available and all samples
with values higher than primary dilution needed to be manu-
ally diluted and afterwards put on analyzers. With TLA, sam-
ples were automatically rerun (for primary and secondary di-
lution) from track-system and sent to analyzers.
Automatic dilution of samples with extreme values in-
creases turnaround-time (TAT), while manual dilution in-
creases TAT and is more vulnerable to errors. Therefore,
manual management of samples with extremely high en-
zyme activities was defined as possible error.
The rate of errors due to manual management was evaluat-
ed and sigma values were calculated. For calculation of sigma
value, total number of tests requested for each enzyme in one
year period was defined as the opportunities, and number of
results above MR for defined enzymes was defined as defects.
Sigma values >3 were defined as acceptable.
Our goal was to reduce number of samples that requited
automatic or manual dilution.
10. Turnaround time (TAT)
Management of samples before and after TLA was differ-
ent. Before TLA emergency samples were centrifuged imme-
diately upon arrival and afterward registered to LIS. Routine
samples were firstly registered to the LIS and afterwards put
on preanalytical system for aliquoting and distribution to other
working places. After TLA implementation, all samples are
registered into the information system immediately after arriv-
al and then put on the track system.
Turnaround time (TAT) was calculated for emergency tests
potassium and troponin in one-month period (November 2013
and March 2016). TAT was measured from registration to the
LIS to releasing results and expressed as median, interquartile
range and 90th percentile of all recorded TAT values (time need-
ed for 90% of samples to be finished). Outliers (number of
samples exceeded 60 and 120 min) were also counted from LIS.
Our goal was to reduce TAT for potassium and troponin in
emergency laboratory.
Page 5 of 12 28
Implementation of TLA
When all 10 KPI’s were measured and recorded in the old
setting, implementation of TLA began. Laboratory space
predefined for new system was cleared out and reconstructed.
Firstly, new analyzers were delivered in the laboratory. Since
Department of Clinical Chemistry is accredited to ISO 15189
standard, all new tests or systems should be verified before
introduction in the routine work. Therefore, verification of
analytical performances (precision, linearity, accuracy, inter-
ferences) for all tests was done. Also, patient’s results were
compared with results from previously used analyzers. If any
bias in reported results were noticed, clinicians were notified
about the possible bias regarding previous results.
When all construction work was finished, analyzers were
connected to the track system. About two months later, when
system was stable and all personnel properly educated, all
defined KPIs were measured again.
Statistical analysis
Data were presented as count and percentages and median
with interquartile range (IQR). Normality of data distribution
was checked with Kolomogorov-Smirnov test. Mann-
Whitney test was used for testing the differences for waiting
time median and turnaround time. Comparison of proportions
was used for percentages of measurement ranges and TAT
outliers before and after implementation of TLA.
Sigma values were calculated using web calculator [7] ac-
cording to the following eqs. [8]:
Defects per million opportunities ðDPMOÞ
¼ ðNumber of defined errors=Total number of opportunities
ðor samplesÞÞ  1000000
Process sigma = 0.8406 + √(29.37)-2.221 × (log(DPMO)).
Sigma values >3 were considered as acceptable values and
P values <0.05 was set as statistically significant. Data analy-
sis was done using MedCalc (MedCalc Software, version
12.5.0.0, Ostend, Belgium).
Results
1. Workspace area
Before TLA implementation, workspace area for the
samples management from entry to LIS to storage
consisted of 4 different distant rooms, separated with
halls. Total area before TLA was 138 m2. After major
construction work and adaptation of workspace, core
28 Page 6 of 12
laboratory space was configured in only two rooms with
total surface of 113 m2 which is almost 20% less space for
preanalytical and analytical sample management and 35%
more space for postanalytical validation, education and
communication with clinicians (Fig. 1). This indicator
was evaluated as successful.
2. LIS registration waiting time
Together with implementation of TLA, the organization
of admission ward was changed and hospital information
system (HIS) was introduced. With HIS, entry of patients
in the LIS was simplified and all data could be transferred
from HIS to LIS using barcode reader.
Waiting time for routine samples was significantly reduced
after TLA implementation and reorganization of admission
with median time 16 (IQR 9–28) before TLA and 9 (IQR 6–
16) minutes after TLA (P < 0.001). Also, 90% of samples
were registered to LIS 13 min earlier after TLA implementa-
tion (43 before vs. 30 min after TLA, respectively). This indi-
cator was evaluated as successful.
3. Tests panel distribution
Before TLA, 13 out of 73 biochemistry tests were mea-
sured on analyzers outside of the core biochemistry analyzer
while after TLA all tests were measured on core analyzers
(78 test; 5 new tests were introduced). Proportion of tests
needed to be manually transferred to other analyzers de-
creased from 1.11% before TLA to 0% after TLA (1301
out of total 117,278 tests performed versus 0 out of
146,234, respectively, P < 0.001). Total number of tests
per month and number of tests per other analyzers before
TLA was expressed as 100%, while values after TLA were
calculated against values before implementation (Fig. 2).
Calculated process sigma value increased from 3.79 to >6.
Fig. 1 Ground plan of the laboratory before and after implementation of total laboratory automation (TLA)
J Med Syst (2018) 42: 28
This indicator was evaluated as successful.
4. Walking distance for sample management
Laboratory personnel members (N = 7) walked approx-
imately 14.3 m per 20 steps (average length of step
0.7 m). Sum of 30 the most common tests performed
in laboratory per day increased for 16% after TLA im-
plementation (4008 before vs. 4662 tests after, respec-
tively). In spite of greater number of tests performed,
number of laboratory workers needed for daily sample
management was lower (9 vs. 8 before and after TLA,
respectively). Moreover, all workers included in the sam-
ple management walked 2 m less per test (5.1 vs. 3.1 m
before and after TLA, respectively), while number of
steps walked by each worker decreased for 31% per
day. When numbers of steps walked per person were
calculated for average of 4662 tests performed daily after
TLA, walking distance decreased for more than 800 m
per person per one day (Fig. 3).
This indicator was evaluated as successful.
5. Tube touch moment (TTM)
6. Manipulation tube time (MTT)
In 2013, before TLA implementation, 20 stages for sample
management from registration to LIS to storage of samples to
refrigerator were possible. The most of the emergency and
routine samples had only tests performed on core biochemis-
try analyzers (91% vs. 98% for emergency and 75% vs 74 for
routine samples before and after TLA, respectively).
Emergency samples before TLA had combination (B) of tests
for aliquoting and core analyzers in 8%. On the other hand,
proportion of routine samples with B combination remained
the same (24%). The smallest proportion (1–2%) of samples
J Med Syst (2018) 42: 28 Page 7 of 12 28

Fig. 2 KPI Test panel distribution

before and after TLA KPI 3. Test panel distribuon
implementation. Total number of 140%
test requests increased for 26%
after TLA with all tests done only 120%
core biochemistry analyzer. 1301
Abbreviations: KPI, key 100%
performance indicator; TLA, total 80%
laboratory automation
60%
40%
20%
0
0%
Total number of tests per month Number of tests per other analyzers
Before TLA Aer TLA

had C combination (only aliquoting tests) before TLA. After
TLA none of emergency sample needed aliquoting for other
analyzers (none with B and C combination). Total tube touch
moment decreased for almost 50% after TLA for emergency
and routine samples. Duration of every step in TTM fell for
approximately 35–40% for routine and emergency samples.
Duration expressed in hours per approximately 500 samples
was lower for more than 4four h per day after TLA introduc-
tion (Fig. 4). Those indicators were evaluated as successful.
7. Analyzers downtime (DT) per month
9. Measurement ranges (MR) and dilution factors (DF)
Values for upper measurement ranges with number of
samples with values above MR during one year are sum-
marized in the Table 2. Sigma values for samples with
possible errors due to manual management (manual load-
ing of samples to the analyzers) was calculated for period
before and after TLA implementation. TLA eliminated
manual manipulation of samples and real sigma values
go to infinity.
This indicator was evaluated as successful.

8. Engagement time (ET) per month 10. Turnaround time (TAT)

DT before TLA was 8875 min (148 h) and after implemen-
tation of TLA 5880 min (98 h). Before TLA analyzers were
out of function even during setting the calibrators and quality
controls on the analyzers. ET decreased after TLA for approx-
imately 28 h per month (118.8 vs. 90.8 h, respectively).
These indicators were evaluated as successful.
Turnaround time for potassium and troponin was longer
after implementation of TLA for approximately 20 min.
Number of samples that exceeded TAT of 1 h was larger 7%
for potassium, and almost 4 times larger for troponin
(Table 3).
This indicator was evaluated as unsuccessful.

Fig. 3 KPI Walking distance for

sample management KPI 4. Walking distance for sample management
140%
4662
120%
9 4008 2639
100% 8
80% 1824
60%
40%
20%
0%
Number of laboratory Number of tests Daily walking distance
staff needed Before TLA Aer TLA per worker (m)
28 Page 8 of 12 J Med Syst (2018) 42: 28

Fig. 4 Tube touch moment

(TTM) and manipulation tube KPI 5. Tube touch moment (TTM) and
time (MTT) calculated before and KPI 6. Manipulaon tube me (MTT)
after implementation of TLA in 17
emergency and routine laboratory 18
16
14 13
12
10 8
7 7.5
8 6.0
6 4.9
3.8
4
2
0
TTM per sample MTT per day (hours) TTM per sample MTT per day (hours)
Emergency laboratory Roune laboratory

Before TLA Aer TLA

Due to the different management of the samples before and
after TLA, new KPI for TAT was defined as sigma values in
which were included number of emergency samples that
exceeded TAT of 1 h per all received samples. As acceptable
criteria sigma value >3 was defined. Sigma values were lower
in the two months period while TLA was introduced. After
that period, sigma values raised to values reached before im-
plementation of the new system (Fig. 5).
Overall, 9 of 10 defined KPIs were successfully measured
and improved after TLA. Only one indicator (median TAT for
potassium and troponin) was not properly defined and could
not be evaluated after implementation of TLA, therefore new
KPIs was established for monitoring TAT.
Discussion
This study showed that total laboratory automation can opti-
mize and diminish many processes in the biochemistry
laboratory, reduce the waste of personnel time and energy
for manual handling of samples and improve entire workflow.
All increment in sample processes efficiency is possible in
even smaller workspace area. Possibilities for errors are sig-
nificantly reduced and personnel are less exposed to possibly
contagious biological material.
If any of the following changes are introduced in the
laboratory: new analytical system, implementation of in-
formational system, reorganization of the workflow, TLA
or accreditation process, it is crucial to asses if those
changes improved efficiency of the processes. KPIs
should be objective and measureable indicators of effi-
ciency in all stages of total testing process (preanalytical,
analytical and postanalytical) [9, 10]. There are several
published studies on using KPIs as a measure of success
in clinical setting, laboratories and clinics. In emergency
departments in Ireland KPIs were defined to monitor clin-
ical outcomes and measure time needed to therapy and to
measure economical cost of collecting data [11]. To

Table 2 Measurement ranges and calculated sigma values for proportion of samples above MR

Enzyme Measurement ranges Percentages of samples above Percentages of samples above Sigma values for Sigma values for
(U/L) MR with automatic dilution MR with manual dilution samples with samples with
automatic dilution manual dilution

Before After Before After P* Before After P* Before After Before After
TLA TLA TLA TLA TLA TLA TLA TLA TLA TLA

Amy (S) 1500 3010 0.33% 0.06% <0.001 0.31% 0.00% <0.001 4.22 4.76↑ 5.34 >6↑↑↑
Amy (U) 4800 3010 3.24% 4.16% 0.022 0.04% 0.00% <0.001 3.35 3.23↓ 4.85 >6↑↑↑
ALT 500 942 0.76% 0.35% <0.001 0.01% 0.00% <0.001 3.93 4.20↑ 5.29 >6↑↑↑
AST 1000 913 0.32% 0.41% 0.003 0.00% 0.00% <0.001 4.23 4.15↓ 5.46 >6↑↑↑
CK 2000 4267 1.70% 0.68% <0.001 0.06% 0.00% <0.001 3.62 3.97↑ 4.73 >6↑↑↑
LD 1200 2000 0.92% 0.39% <0.001 0.01% 0.00% <0.001 3.86 4.18↑ 5.32 >6↑↑↑

Abbreviations: Amy (S), serum amylase; Amy (U), urine amylase; ALT, alanine aminotranspherase; AST, aspartate aminotranspherase; CK, creatine
kinase; LD, lactate dehydrogenase; MR, measurement range. *Statistically significant values are bold. Arrows represent decrease or increase of sigma
values.
J Med Syst (2018) 42: 28 Page 9 of 12 28

Table 3 Turnaround time for

potassium and troponin in Before TLA (October 2013) After TLA (March 2016) Change P*
emergency laboratory before and
after implementation of the TLA Potassium N = 6866 N = 7677
expressed in minutes and as
percentages of outliers Median, IQR (min) 29 (20–57) 45 (38–69) ↑ 66% <0.001
Outliers 1 h (%) 25.8 27.6 ↑ 7% 0.015
Outliers 2 h (%) 2.8 3.4 ↑ 21% 0.042
Troponin N = 546 N = 1139
Median, IQR (min) 24 (19–31) 46 (40–52) ↑ 100% <0.001
Outliers 1 h (%) 3.3 12 ↑ 363% <0.001
Outliers 2 h (%) 0.4 0.3 ↓ 25% 0.906

Abbreviations: IQR, interquartile range; TLA, total laboratory automation. *Statistically significant values are
bold.

estimate efficiency in public laboratory network in Brazil,
group of researchers defined several indicators which in-
cluded characteristics of tests produced, usage of tests per
patient and cost of personnel, tests and equipment used in
the laboratory network [12]. Although recently different
laboratories measured miscellaneous indicators, none of
them was completely applicable to our setting. Every lab-
oratory has its own specificities in the workflow, sample
management and organization of other processes.
Therefore, it is necessary to define own specific and mea-
surable criteria of improvement. KPIs in our laboratory
were defined before TLA implementation for the most
erroneous and possible vulnerable processes. Our KPIs
were not only technical (e.g. acquisition of new equipment
and adaptation of workspace area), but also organizational.
Changes in organization of processes were performed dur-
ing and after implementation in order to optimize as much
processes as possible.
Dimension of the workspace depends on assigned area for
the laboratory in the hospital and usually it is impossible to
enlarge it. However, reorganization of the workflow, new and
modern analyzers that are usually smaller than the old ones,
make possible to reduce space for analytical and preanalytical
procedures. The worldwide request for better space utilization
of laboratory was previously recognized. Automation brings
cost reduction as well as quality improvement [13].
Implementation of TLA in our laboratory significantly re-
duced area needed for analytical and preanalytical sample
processing, leaving larger area for postanalytical procedures.
In the past years, analytical systems are highly sophisticated
with very low error rate [14]. Also, reducing errors in
preanalytical phase has been put in great focus [15].
However, great attention should be paid on postanalytical
phase, especially on interpretation of results, reporting critical
results and improving patients healthcare with reducing pos-
sible errors in diagnosis [16–19]. With new organization of
our workspace, specialists in laboratory medicine have larger
area for postanalytical analysis of results, communication with
clinicians and scientific work.
Although modification of registration of samples to
LIS was purely organizational and not dependent on
TLA implementation, KPI showed significant reduction
(50%) of wasted time for samples. With new organization,
samples are processed sooner, stability of tests in speci-
mens is not compromised and possibility of errors in sam-
ple management is reduced [20].

Fig. 5 Sigma values for samples Sigma

that exceeded 60 min TAT. Arrow values

shows the month in which TLA 4.0

was implemented 3.5
3.0
2.5
2.0
1.5
1.0
0.5
Year_month

0.0
28 Page 10 of 12
The most important benefit of TLA is reduction of the
probability for errors which was very high with manual
handling of samples [21].Especially vulnerable to errors
are processes such as labelling of aliquot tube and man-
ual dilution of samples with extreme values. Errors in
labeling could lead to severe risk for patient’s safety
and healthcare quality [22]. In our laboratory number
of samples needed to be manually handled was mini-
mized with TLA, therefore the potential errors were also
reduced. Our laboratory is Reference center for
Emergency laboratory medicine of the Ministry of
Health, Republic of Croatia, and has many pathological
samples, thus is particularly important to have wide mea-
surement ranges at analyzers and automatic dilution for
the highest test results. TLA enables double and even
triple rerun of samples and automatic dilution of extreme
values. With automated procedures, dilution errors as a
result of human mistakes are almost non-existing. Not
only that TLA decreases possibility for errors but with
properly planned dilution management also TAT can be
significantly reduced. In the study of Ialongo et al. the
TAT for troponin was reduced for almost 80% for
prediluted samples of patients with suspected high values
of troponin [23].
Furthermore, implementation of TLA significantly re-
duced number of tube touches needed for transferring the
samples to analyzers other than core biochemistry ana-
lyzer. With all those improvements in sample processing,
exposition of personnel to potentially contaminated bio-
logical material is considerably reduced with TLA. In the
previous study we have confirmed that personnel usually
do not comply with recommended safety measures [24],
therefore we can conclude that TLA may contribute to
safety of personnel and consequently improve quality of
results with positive impact on patient healthcare.
With introduction of automation time used for sample
management and analyzer maintenance could be substan-
tially saved. Decreased energy waste for unneeded proce-
dures and reduction of walking distance for sample man-
agement allows higher rate of productivity and efficiency
for staff members without compromising patient safety
[25]. Besides that, contact with possible contaminated
biological material is reduced and procedures are stan-
dardized. Our data collected for only two years after im-
plementation of the TLA are concordant to the results
provided in the long term study of Sarkozi et al. They
collected data for 36 years and proved that automation
increase productivity of personnel, while number of em-
ployees and price per reported test could be reduced [26].
The most of defined and measured KPIs in this study
showed significant improvement of procedures after
TLA implementation, excluding the turnaround time.
TAT is highly important for clinicians and clinical
J Med Syst (2018) 42: 28
decisions for critically ill patients [27]. Median TAT for
potassium and troponin after TLA implementation in our
study rose to 45 and 46 min, respectively. Our results are
similar to previously published study of Angeletti et al.
where median TAT for K and troponin in their laboratory
in Italy, was 39 and 42 min, respectively [28]. Likewise,
Lou et al., in Canada, measured 34 min as median TAT
for potassium [29]. However, although their median TAT
was slightly lower than ours, those values are not
completely comparable to ours, especially due to differ-
ent percentages of emergency samples and number and
type of analyzers included in the TLA. Our laboratory
has almost 50% of emergency samples while Canadian
and Italian have only 30% and 3%, respectively.
Moreover, laboratory in Canada has 3 biochemistry ana-
lyzers (Architect c16000) and 5 immunochemistry ana-
lyzers (Architect i2000) connected to TLA while our lab-
oratory has two slower Architect c8000 and only one
Architect i2000. Other laboratory in the Italy, has differ-
ent analyzers connected to TLA, so those results are not
entirely comparable to ours. Nevertheless, recently pub-
lished extensive research in almost 480 laboratories ob-
tained median TAT for potassium of 43 min which is
similar to our results [30].
Our process from sample registration to LIS to putting
the samples on analyzers was reorganized and true value
for TAT was not possible to calculate. Before TLA, the
most of the emergency samples were centrifuged before
registration to LIS. In the meantime HIS was introduced
which increased speed of sample registration to LIS and
samples are now immediately put on the TLA. This
brings the difference for TAT which was not measurable
or comparable. Because of that we have introduced new
KPI, number of outlier samples with TAT exceeding de-
sirable 60 min for reporting results [31]. Those outliers
were expressed as sigma values which were lower than
predefined 3 only two months after TLA implementation,
which can be considered as time for adaptation to new
system and process organization. After initial period, sig-
ma values raised and now are constantly above 3.
This study has some limitations. Our defined KPIs are
specific for our workflow and facilities and are probably
not entirely applicable to other laboratories with different
s a m p l e m a n a g e m e n t . M o r e o v e r, d u e t o r i s e o f
informatization level in hospital and reorganization of
processes some predefined KPIs were not measurable
or comparable with values gained after implementation
of the TLA, e.g. TAT. Although it is important to define
KPIs that are measurable, easy to follow, representative
and time independent [5, 11], it is not easy to define
objective measures before new system is introduced.
The system is constantly changing and KPIs or other
measures of improvement should be changed
J Med Syst (2018) 42: 28 Page 11 of 12 28

accordingly in order to describe real working conditions.
Every laboratory should define their own KPIs and our
KPIs may only be template for that.
To define their own KPIs before implementation of new
system, every laboratory can follow some recommendations
that arose from our study and are given in flowchart (Fig. 6).
In the conclusion, implementation of the total labora-
tory automation improved many processes in our labo-
ratory. TLA reduced waiting time for samples registra-
tion, wasting time of personnel for manual sample han-
dling and analyzers maintenance as well as contact of
personnel with contaminated biological samples.

Fig. 6 Flowchart of
recommended steps for
implementation of changes in
laboratory with explained steps
on the left side of the chart
28 Page 12 of 12
Funding This research received no specific grant from any funding agen-
cy in the public, commercial, or not-for-profit sectors.
Compliance with Ethical Standards
Conflict of Interest Author Marijana Miler declares that she has no
conflict of interest. Author Nora Nikolac declares that she has no conflict
of interest. Author Lora Dukic declares that she has no conflict of interest.
Author Ana-Maria Simundic declares that she has no conflict of interest.
Ethical Approval Not applicable.
Human and Animal Rights and Informed Consent This article does not
contain any studies with human participants or animals performed by any
of the authors.
References
1. Hawker, C.D., Laboratory automation: total and subtotal. Clin. Lab.
Med. 27(4):749–770, 2007.
2. Zaninotto, M., and Plebani, M., The Bhospital central laboratory^:
automation, integration and clinical usefulness. Clin. Chem. Lab.
Med. 48(7):911–917, 2010.
3. Cameron, P.A., Schull, M.J., and Cooke, M.W., A framework for
measuring quality in the emergency department. Emerg. Med. J. 28:
735–740, 2011.
4. Harvey, H.B., Hassanzadeh, E., Aran, S., Rosenthal, D.I., Thrall,
J.H., and Abujudeh, H.H., Key performance indicators in radiolo-
gy: you can’t manage what you can’t measure. Curr. Probl. Diagn.
Radiol. 45(2):115–121, 2016.
5. Bovend’Eerdt, T.J.H., Writing SMART rehabilitation goals and
achieving goal attainment scaling: a practical guide. Clin.
Rehabil. 23:4352–4361, 2009.
6. Armbruster, D.A., Overcash, D.R., and Reyes, J., Clinical
Chemistry Laboratory Automation in the 21st Century - Amat
Victoria curam (Victory loves careful preparation). Clin. Biochem.
Rev. 35(3):143–153, 2014.
7. https://www.isixsigma.com/process-sigma-calculator/. Accessed 8
Apr 2017.
8. https://www.easycalculation.com/statistics/learn-process-sigma.
php. Accessed 8 Apr 2017.
9. Rizk, M.M., Zaki, A., Hossam, N., and Aboul-Ela, Y., Evaluating
laboratory key performance using quality indicators in Alexandria
University Hospital Clinical Chemistry Laboratories. J. Egypt Publ.
Health Assoc. 89(3):105–113, 2014.
10. Salinas, M., López-Garrigós, M., Gutiérrez, M., Lugo, J., Sirvent,
J.V., and Uris, J., Achieving continuous improvement in laboratory
organization through performance measurements: a seven-year ex-
perience. Clin. Chem. Lab. Med. 48(1):57–61, 2010.
11. Gannon, B., Jones, C., McCabe, A., O’Sullivan, R., and Wakai, A.,
An economic cost analysis of emergency department key perfor-
mance indicators in Ireland. Eur J Emerg Med. 24(3):196–201, 2017.
12. Santos, M.A., Moraes, R.M., and Passos, S.R., Performance indi-
cators and decision making for outsourcing public health laboratory
services. Rev. Saude Publica. 46(3):456–465, 2012.
13. Markin, R.S., and Whalen, S.A., Laboratory automation: trajectory,
technology, and tactics. Clin. Chem. 46(5):764–771, 2000.
14. Plebani, M., Errors in clinical laboratories or errors in laboratory
medicine? Clin. Chem. Lab. Med. 44:750–759, 2006.
J Med Syst (2018) 42: 28
15. Simundic, A.M., and Lippi, G., Preanalytical phase – a continuous
challenge for laboratory professionals. Biochem. Med. (Zagreb).
22(2):145–149, 2012.
16. Kopcinovic, L.M., Trifunović, J., Pavosevic, T., and Nikolac, N.,
Croatian survey on critical results reporting. Biochem. Med.
(Zagreb). 25(2):193–202, 2015.
17. Favaloro, E.J., Lippi, G., and Adcock, D.M., Preanalytical and
postanalytical variables: the leading causes of diagnostic error in
hemostasis? Semin. Thromb. Hemost. 34(7):612–634, 2008.
18. Katz, C., McNicholas, K., Bounds, R., Figurelle, T., Jones, C.,
Farley, H., Witkin, G., McLane, M.A., and Johnson, S.R.,
Improving patient safety through enhanced communication be-
tween emergency department clinicians and medical laboratory
staff. J. Clin. Outcomes Manag. 20(10):455–462, 2013.
19. Flegar-Meštrić, Z., Perkov, S., and Radeljak, A., Standardization in
laboratory medicine: Adoption of common reference intervals to
the Croatian population. World J. Method. 6(1):93–100, 2016.
20. Cuhadar, S., Atay, A., Koseoglu, M., Dirican, A., and Hur, A.,
Stability studies of common biochemical analytes in serum separa-
tor tubes with or without gel barrier subjected to various storage
conditions. Biochem. Med. (Zagreb). 22(2):202–214, 2012.
21. Lam, C.W., and Jacob, E., Implementing a laboratory automation
system: experience of a large clinical laboratory. J. Lab. Autom.
17(1):16–23, 2012.
22. Martin, H., Metcalfe, S., and Whichello, R., Specimen labeling
errors: a retrospective study. Online J. Nurs. Inform. 19(2), 2015.
Available at http://www.himss.org/ojni. Accessed 22 May 2017.
23. Ialongo, C., Pieri, M., and Bernardini, S., Artificial neural network
for total laboratory automation to improve the management of sam-
ple dilution. LAS Technol. 22(1):44–49, 2017.
24. Dukic, K., Zoric, M., Starcic, J., Culjak, M., Saracevic, A., and
Miler, M., How compliant are technicians with universal safety
measures in medical laboratories in Croatia? Biochem. Med.
25(3):386–392, 2015.
25. Miler, M., and Nikolac, N., Patient safety is not compromised by
excluding microscopic examination of negative urine dipstick. Ann.
Clin. Biochem., 2017. https://doi.org/10.1177/0004563216687589.
26. Sarkozi, L., Simson, E., and Ramanathan, L., The effects of total
laboratory automation on the management of a clinical chemistry
laboratory. Retrospective analysis of 36 years. Clin. Chim. Acta.
329(1–2):89–94, 2003.
27. Goswani, B., Singh, B., Chawla, R., Gupta, V.K., and Mallika, V.,
Turn around Time (TAT) as a Benchmark of Laboratory
Performance. Indian J. Clin. Biochem. 25:376–379, 2010.
28. Angeletti, S., De Cesaris, M., Hart, J.G., Urbano, M., Vitali, M.A.,
Fragliasso, F., and Dicuonzo, G., Laboratory automation and intra-
laboratory turnaround time: experience at the University Hospital
Campus Bio-Medico of Rome. J. Lab. Autom. 20(6):652–658,
2015.
29. Lou, A.H., Elnenaei, M.O., Sadek, I., Thompson, S., Crocker, B.D.,
and Nassar, B., Evaluation of the impact of a total automation sys-
tem in a large core laboratory on turnaround time. Clin. Biochem.
49(16–17):1254–1258, 2016.
30. Fei, Y., Zeng, R., Wang, W., He, F., Zhong, K., and Wang, Z.,
National survey on intra-laboratory turnaround time for some most
common routine and stat laboratory analyses in 479 laboratories in
China. Biochem. Med. (Zagreb). 25(2):213–221, 2015.
31. Pati, H.P., and Singh, G., Turnaround Time (TAT): difference in
concept for laboratory and clinician. Indian J. Hematol. Blood.
Transfus. 30(2):81–84, 2014.