International Journal of Laboratory Hematology

The Official journal of the International Society for Laboratory Hematology

ORIGINAL ARTICLE INTERNAT IONAL JOURNAL OF LABORATO RY HEMATO LOGY

Full Blood Count – Internal QC Protocol: a review by the Royal

College of Pathologists of Australasia Quality Assurance Programs
(RCPAQAP) Pty Ltd – Haematology
J. SIOUFI*, T. BADRICK*, L. SINCLAIR † , K. MARSDEN*

*RCPAQAP Haematology, St. S U M M A RY

Leonards, NSW, Australia
†
Sullican Nicolaides Pathology, Introduction: The RCPAQAP (Royal College of Pathologists of Aus-
Wesley Private Hospital, tralasia Quality Assurance Program) Haematology has undertaken
Auchenflower, QLD, Australia
an exercise to review the internal quality control protocol for full
Correspondence: blood count (FBC) instrumentation as well as review the action
John Sioufi, Royal College of taken by laboratories when nonconforming results are evident in
Pathologists Quality Assurance the RCPAQAP proficiency testing reports.
Programs Pty Ltd, Suite 201,
level 2, 8 Herbert Street, St.
Method: A questionnaire was sent to laboratories enrolled in the
Leonards, Sydney, NSW, RCPAQAP FBC module. Laboratories were asked to provide infor-
Australia. mation with regard to the type of control measures used within
Tel.:+61 2 9045 6040; their testing environment that would trigger alerts for nonconform-
Fax: +61 2 9438 5378;
E-mail: john.sioufi@ ing events. The questionnaire also reviewed the action taken by
rcpaqap.com.au laboratories in response to nonconforming test results in their
external QA reports.
Results: A total of 253 of the 850 laboratories enrolled in the FBC
doi:10.1111/ijlh.12582
module returned a response to the questionnaire, which identified
variation in the QC protocol used to identify nonconforming events
Received 22 April 2016;
accepted for publication 17 on the FBC analyser, including the type of controls, control levels
August 2016 processed and the frequency of use.
Conclusion: This questionnaire identified variation in the internal
Keywords QC protocol used by laboratories, including the types of control
RCPAQAP, FBC, quality, measures used and the rules applied to identify nonconforming
internal, external, protocol
events. However, the majority of laboratories appear to follow the
most favourable choice of actions supplied when reviewing results
of external QA data.

Australia and overseas. Over 800 laboratories partici-

INTRODUCTION
pate in the RCPAQAP – Haematology surveys for
The RCPAQAP(Royal College of Pathologists of Aus- the full blood count (FBC).
tralasia Quality Assurance Program) provides exter- Internal quality control (IQC) and proficiency test-
nal quality assurance/proficiency testing in all ing/external quality assurance (PT/EQA) play a vital
pathology specialties to over 1000 laboratories in role in the accuracy and precision of test results

84 © 2016 John Wiley & Sons Ltd, Int. Jnl. Lab. Hem. 2017, 39, 84–94
 J. SIOUFI ET AL. | QC PROTOCOL FOR FBC INSTRUMENTATION 85

produced in pathology laboratories. The FBC test, Addition, which states ‘the value of an automated
sometimes referred to as the complete blood count hematology analyzer is to provide physicians and other
(CBC), is a commonly requested blood test and with health care providers with reliable hematology data for
automated analysers now includes white cells, red patient management. From a patient and regulatory per-
cells, haemoglobin, platelets and other parameters for spective, all complete blood count (CBC) results should be
further characterization of these cells. statistically and medically comparable on any hematology
A variety of manufacturers produce FBC instrumen- analyser’ [1]. Interpretation of these standards can vary
tation, and these use several different principles to amongst laboratories, therefore we undertook a study to
obtain measurements for the FBC parameters. For exam- determine the practices of the large group of laboratories
ple, there are instruments that measure the white cell who participate in the RCPAQAP Haematology - FBC EQA
count by impedance counting, laser light scatter or flow program.
cytometry. Even though these principles of measure-
ment are different, patient test results should yield the
Purpose
same result from instrument to instrument, no matter
what principle is being used to derive the measurement. The aim of this study was to determine and evaluate the
Manufacturers supply laboratories with mainte- internal QC protocols used by laboratories when pro-
nance and quality assurance protocols, specific for the cessing FBC and to review the action taken by laborato-
models purchased, with the aim of ensuring that the ries in response to nonconforming results in their
instrument is operating efficiently at all times. The RCPAQAP PT/EQA reports. We also hope to assist labo-
maintenance of any instrument that is frequently used ratories in best practice when performing internal QC as
should be performed on an at least daily basis to con- well as suggest appropriate action to take when external
tinually monitor the operations and controls run to QA reports are nonconforming when compared to their
check the data produced. It also enables the operator peer group.
to detect any problem that could be occurring, using
some added internal quality measures to ensure pre-
M AT E R I A L S A N D M E T H O D S
cise test results. The IQC samples should cover a
broad range of levels for each test parameter pro- A questionnaire was developed by the RCPAQAP to
cessed on the FBC instrument. gather information about the IQC used by participants
In conjunction with IQC measures, laboratories also for their FBC instrumentation, including the practices
participate in PT/EQA programs, a process whereby the used for multiple instruments and for out-of-hours ser-
quality of laboratory results can be confirmed by peer vices with the intention of providing peer review and
group comparison, ensuring the accuracy of test results. guidance for good laboratory practice. This question-
Laboratories will then review results for trends or naire focussed only on internal QC during the actual
biases, based on peer group comparison, to determine analytical process of FBC testing and did not cover the
any issues that may be occurring within their testing pre- and postanalytical aspects which should also be
environment. part of the total quality control process. Other questions
Internal quality control and EQA are usually depen- aimed to determine laboratory practice in reviewing
dent on stabilized blood, which has been treated with EQA results and the action taken when results were
external agents to prolong the life of the sample. These outside the acceptable range.
additives usually affect the constitution or matrix of the The questionnaire was split into two parts: Part A
blood cells and can cause a difference in readings when asked questions about the Internal QC practice used
compared to patient material. Internal quality control in laboratories and Part B was to assess laboratory
methods also use statistical analysis to review the mean practice when RCPAQAP survey results in the FBC
of their results and to ensure the reliability of test results module were flagged for review.
by reviewing trends or biases over a set period of time. The questions in Part A related to the number of
CSLI have produced standards, H26-A2 – Validation, instruments that were in operation in laboratories and
Verification and Quality Assurance of Automated the internal QC protocol for each instrument, the type
Hematology Analyzers; Approved Standard – Second of quality control material used to monitor each

© 2016 John Wiley & Sons Ltd, Int. Jnl. Lab. Hem. 2017, 39, 84–94
86 J. SIOUFI ET AL. | QC PROTOCOL FOR FBC INSTRUMENTATION

instrument, the use of the inbuilt facility on FBC Question 2: How many FBC samples are processed
instrumentation for monitoring the performance and through your instrument on a daily basis? Laborato-
the rules applied by laboratories to highlight noncon- ries with many instruments were asked to return their
forming events. response based on the most frequently used instru-
The questions in Part B aimed to determine ments, or the first 3 instruments in their laboratory.
whether correct laboratory practice was applied when (Figure 2).
results from the external QA survey reports for the A large number of laboratories use more than one
FBC module were flagged for review in three different instrument. Laboratories using more than three
scenarios. instruments in the same location were asked to pro-
The questionnaire was sent online to 850 labora- vide information only for the three most frequently
tories enrolled in the RCPAQAP Haematology FBC used instruments in their laboratory.
module. All results were analysed using Microsoft Some laboratories that have more than one FBC
Excel. instrument run patient samples alternately on each
instrument, thus having the same daily throughput on
all their instruments while others will only use the sec-
R E S U LT S
ond or third instrument as a backup, thus processing
A response was received from 253 (29.6%) of the 856 lower numbers of samples on the backup instruments.
laboratories that were sent the questionnaire, and the
questions and responses are listed below.
Participants with more than one FBC instrument in their
laboratory were asked to base the following questions
Part A: internal QC questions on their primary instrument

Question 1: What make and models of FBC instruments Question 3: What types of internal controls are used on
are used in your laboratory? (e.g. Sysmex XN10, Cell your FBC instrument?
Dyn 3200 etc.). See Figure 1.

Instrumentation
60

50

40

30

20

10

Figure 1. Illustrates the FBC instrumentation used by laboratories, which covered a wide range of instrument model
types. Even though different principles are used to derive FBC parameters, the internal quality control measures
used by laboratories should be similar.

© 2016 John Wiley & Sons Ltd, Int. Jnl. Lab. Hem. 2017, 39, 84–94
 J. SIOUFI ET AL. | QC PROTOCOL FOR FBC INSTRUMENTATION 87

Number of samples processed daily

100

90

80

70
No. Participants

60

50

40

30

20

10

0
<50 <100 <200 <300 <500 >500
No. Samples

Figure 2. Number of samples processed daily.

Of the 253 responding laboratories, 234 (92.5%) that run their tri-level controls every 24 h stated that
laboratories used controls specific for the FBC instru- drift controls are run more regularly (every 4 h).
ment/model in use in their laboratory. Question 5: Do you have different criteria for after-
Thirteen of the 234 laboratories (5.5%) also pur- hours work?
chased third-party controls. Eleven laboratories The majority of laboratories 188/245 (78%)
(4.7%) only used commercial samples from a third reported that the same rules applied for after-hours
party. work. From the 22% (57/245) of laboratories that
A total of 120 of the 253 laboratories (47.4%) used use different protocols after hours, 36/57 laboratories
drift controls in conjunction with the internal control (63%) indicated that the normal-level or drift con-
material purchased from a manufacturer or third trols (commercial or in-house) were processed
party. after hours. The remaining laboratories that
Question 4: What control levels are used and how responded to this question indicated either that
often are they run. they had an on-call service and ran all three
The majority of participants who responded to levels or ran only one of the tri-levels with
this question 242/244 (99%) processed high, low instrument start-up – 7/57 (12%). One laboratory
and normal controls through their FBC instrumenta- stated that no QC was run between 24:00 h and
tion. 07:00 h.
Two laboratories ran ‘normal’-level controls only, Question 6: Do you process your internal QC sam-
both laboratories stating that ‘drift controls are run in ples ‘all at the one time (high, low and normal con-
conjunction with the normal control’. (Table 1). trols each time you perform internal QC)’ or ‘normal
From the results above, the most common practice control in morning, low control during the day and
was to run tri-level controls every 24 h. There does high control in evening’?
not seem to be a pattern related to the number of From the 211 laboratories that responded to this
samples processed per day. A group of laboratories question, 197 laboratories (93%) processed their

© 2016 John Wiley & Sons Ltd, Int. Jnl. Lab. Hem. 2017, 39, 84–94
88 J. SIOUFI ET AL. | QC PROTOCOL FOR FBC INSTRUMENTATION

Table 1. Frequency of processing internal QC samples within laboratories

No. of Samples being

processed (number Every Every Every Morning, day &
of laboratories) 4h 8h 24 h Evening Other

<50 (56 laboratories) 3 13 28 6 6*

<100 (61 laboratories) 7 14 30 3 7†
<200 (47 laboratories) 8 11 19 4 5‡
<300 (31 laboratories) 2 8 13 4 4§
<500 (36 laboratories) 4 6 14 4 8¶
>500 (23 laboratories) 1 5 11 2 4**

Please note: Only the primary instrument was used to ascertain how frequently the internal controls were being pro-
cessed.
’Other’: Controls were run at intervals other than those stated.
*One laboratory stated every 12 h, three laboratories stated they run tri-levels in the morning then alternating levels
in the afternoon, one laboratory stated that they run known patient samples every day and commercial controls every
3 months. One laboratory stated that they run controls once a day three times per week.
†
Two laboratories stated every 12 h, two laboratories run tri-levels in the morning then alternating levels in the after-
noon, two laboratories stated every 24 h with a drift control every 4 h and one laboratory runs one of each tri-level at
intervals during the day.
‡
Two laboratories stated every 12 h, one laboratory runs tri-levels once a day and then one of each level every 8 h,
one laboratory runs tri-levels every 4 h during the day and then one level after hours. One laboratory runs tri-levels at
each start-up.
§
One laboratory runs tri-levels every 6 h and three laboratories run tri-levels morning and day. Two laboratories run
tri-levels every 24 h and drift controls every 4 h.
¶
Three laboratories stated every 12 h, one laboratory runs tri-level controls every 3 h, one laboratory runs tri-levels
once every 24 h then the normal control every 4 h. Two laboratories run tri-levels every 24 h and drift controls every
4 h, one laboratory runs tri-levels every 80–100 samples.
**One laboratory runs tri-levels every 10 h, one laboratory runs each tri-level control once during the day, one labora-
tory runs the drift every 4 h and the commercial tri-level in the morning and evening, and one laboratory ran a high
or normal control every 8 h.

high, normal and low controls at the same time 2. Sudden change in results from the mean, either
when performing internal QC. positive or negative.
3. A gradual drift of any parameter.
Some laboratories that processed all tri-levels at each
4. Results that were greater than 2SD above or
episode stated that they alternate between open and
below the mean.
closed modes.
5. Results that were greater than 3SD above or
Laboratories that stated ‘OTHER’ process all levels
below the mean.
once and then alternate between levels at intervals
6. One result >3SD from the mean, three results
during the 24 h.
>2SD from the mean or 10 sequential results on the
Question 7: Participants were asked to note 3–5 same side of the mean.
examples of what they would consider non-con- 7. Ongoing bias of results, continually above or
forming Internal QC results when running H/N/L below the median.
controls. 8. Westgard rule violations (1 2s, 1 3s, 2 2s, R4s, 3
Listed below are some of the rules reported: 1s, 4 1s)
9. Systematic error, when more than one control
1. One or more results that were outside the exceeded the same limits.
allowable limits of performance set by the manufac- 10. Random error, when a single result was outside
turer. manufacturer limits.

© 2016 John Wiley & Sons Ltd, Int. Jnl. Lab. Hem. 2017, 39, 84–94
 J. SIOUFI ET AL. | QC PROTOCOL FOR FBC INSTRUMENTATION 89

Question 8: Do you run a drift control? (mixing/sampling) – 43.0%’, followed by ‘QC material
Of the 230 laboratories that responded to this ques- integrity’ – 39.5% and 6.6% stated an ‘analyser drift or
tion, 149 (65%) stated that a drift control was run with failure’. The remaining 10.9% of comments included
the tri-level controls purchased from the manufacturer low volume of sample, inconsistent power supply, sam-
or third party. ple chosen has low parameters or short sampling.
A total of 88 of the 149 (59%) laboratories stated that Question 11: As part of your internal QC do you
the source of the drift control was commercial material, review the ‘mean of patient results’, i.e. moving aver-
and 60 (40%) laboratories used patient material. age or XB analysis?
One laboratory used both patient and commercially Of the 223 laboratories that responded to this ques-
prepared material as a drift control. tion, 162 (72.6%) review the mean of patient results,
Laboratories were also asked how often drift controls and the remaining 61 (27.4%) laboratories indicated
were run in their laboratory, with 98 participants that this review was not performed.
(42.5%) indicating every 4 h, 22 (9.5%) ran drift con- Question 12: Participants were asked when review-
trols every 8 h and seven (3.0%) ran drift controls every ing the ‘mean of patient results’ to state 3 main indi-
12 h. Other responses included running drift controls cators of analyser issues.
with every batch of patient samples – 25 (11.0%), every A total of 138 laboratories responded to this ques-
2 h, 39 (17.0%) and every 3 h seven (5.0%). There tion, with the five most frequently used indicators of
were also a variety of other responses including every analyser issues below:
24 h, every hour or random processing.
1. Out-of-range results for the MCV, MCH or MCHC
Question 9: If you do run drift controls, how are
(most common stated) or Hb, RCC, Hct and platelets
allowable limits set?
also noted – 55 (40.0%).
Below are the most frequently used rules reported
2. The trend or drift in any parameter – 28 (20.0%)
by the 137 laboratories that responded to this ques-
3. Outside of range set for any parameter 17
tion:
(12.0%)
1. Limits set by the manufacturer when commercial 4. Two or more consecutive alarms 14 (10%)
control was used – 47 (35.0%). 5. Set absolute ranges for parameters (5% range for
2. Based on the Commercial Control limits – 14 RCC) or trends above or below 2 or 3SD – 12 (9.0%).
(10.0%).
Question 13: When do you calibrate your instrument?
3. Establishing the average and then 2SD range – 22
Of the 216 laboratories that responded, the most com-
(16.0%). The number of times samples were processed
mon interval was at 6 months – 94 (43.5%), followed by
to establish the average ranged from 2 to 20 times, a
three monthly – 30 (13.9%), with 11 (5.6%) starting that
minority group doing repeatability testing on all their
they calibrate their instruments at yearly intervals.
instruments to establish the average.
Laboratories also stated that calibration of instrumen-
4. Established by the Central/Main laboratory – 10
tation was performed only when required 61 (28.2%) or
(7.5%)
after any major repair or as needed – 10 (4.6%).
5. Set automatically by the instrument – 6(4.5%)
A few laboratories were not sure when calibration
6. Set ranges dependent on the parameter, for example
was performed or depended on the manufacturer to
10% range for WCC, 15% range for Platelet count – 4
perform the calibration – 4 (2%).
(3.0%)
A minor group of four laboratories (2%) performed
Some other rules mentioned were setting accept- calibration when significant errors were seen in their
able limits based on biological variation of each Internal QC or external QA data.
parameter, RCPAQAP limits used, Levey–Jennings One laboratory stated that they never calibrated
plots for obvious drifts and limits based on CV%. their instrument, and another indicated that this was
Question 10: What is the most frequent cause of rarely required for their Sysmex XS instrument.
drift control outliers? Question 14: Laboratories were asked if instrument
The two most frequent causes of drift control outliers to instrument comparisons were performed if more
reported by laboratories were ‘inappropriate handling than 1 instrument was located on sight.

© 2016 John Wiley & Sons Ltd, Int. Jnl. Lab. Hem. 2017, 39, 84–94
90 J. SIOUFI ET AL. | QC PROTOCOL FOR FBC INSTRUMENTATION

A total of 55 laboratories responded to this ques- of performance in your External QA survey report?
tion, with 45 laboratories (82%) stating that instru- See Figure 3.
ment-to-instrument comparison is performed. The Question 2: What are your next steps if the WCC,
majority of these laboratories have more than one Hb or Platelets are consistently high or low in 3 con-
instrument on-site or perform instrument-to-instru- secutive survey reports?
ment comparison at another site. Five of the 45 lab- From the 207 responses received, 183 (88.5%)
oratories stated that they perform this regularly on responded by saying they would ‘check to see if a
either a daily or weekly basis. Ten of the 55 labora- similar pattern is occurring with internal QC data,
tories (18%) that responded stated that no instru- and then would consider to recalibrate’, which is
ment-to-instrument comparison was performed. considered the recommended corrective action.
Nineteen laboratories (9.1%) would recalibrate their
instrument and five laboratories (2.4%) would check
Part B: action taken when reviewing external quality
their internal QC data on the day of testing, and if
assurance results
it was in the acceptable range, they would do noth-
The following questions were asked to ascertain the ing.
actions taken by laboratories if their EQA results were A small group stated that they would not wait for
nonconforming. three consecutive failures and would consider investigat-
Question 1: What would be the next step taken If ing trends from other users with the same
the WCC, Hb or Platelets were outside allowable limits instrument.

180

160 72.6%

140

120
No. Participants

100

80

60
24.1%

40

20
1.9% 1.4%
0
Check your internal QC Do nothing and wait for Recalibrate your Repeat testing of the
data on the day of testing the next report to see whether instrument survey samples
the results come back in

Figure 3. Illustrates that the majority of participants stated that they would check their Internal QC on the day of
testing for similar trends. One laboratory also stated that they would check previous internal QC and external QAP
results to see whether this was trending prior to the issue, which is good laboratory practice.

© 2016 John Wiley & Sons Ltd, Int. Jnl. Lab. Hem. 2017, 39, 84–94
 J. SIOUFI ET AL. | QC PROTOCOL FOR FBC INSTRUMENTATION 91

200
92.9%
180

160

140

120

100

80

60

40

20 3.0% 3.0%
1.0%
0
Check the End-of-Cycle report Check your internal QC for your Do nothing, results are still Recalibrate your instrument
ranking and if you are ranked at high controls to see whether within acceptable limits
50% compared to all other a positive bias is seen, that is
participants do nothing Internal QC results always higher
than the target, consider recalibration

Figure 4. Responses from participants when reviewing nonlinear patterns of results in the RCPAQAP End-of-Cycle
reports.

Question 3: Laboratories were asked what they results or to detect any nonconforming test results
would do if they reviewed their linearity charts from that may occur in normal operations. The quality con-
the End of Cycle report and there appeared to be a trol process within a laboratory environment is also
non-linear pattern occurring with their platelet essential to verify and validate test results before they
results, i.e. high platelet counts appear to be high, are released. Therefore, it is the responsibility of the
normal and low platelet counts closer to the median. laboratory to establish an IQC process that will moni-
(Figure 4). tor the entire analytical process, such as running con-
The majority of respondents selected the most trol samples and calibrators of known values on a
appropriate response, which was to check their inter- regular basis, depending on how frequently samples
nal QC for the high controls to see whether a positive are run through the instrument. Each FBC instrument
bias was seen. A few laboratories simply stated that produced in the market will vary in the principle used
they would liaise with the instrument representative to generate blood cell parameters; however, standard
to correct the issue. approaches can be taken to establish IQC checks. This
Even though ranking appears acceptable and questionnaire aimed to determine the degree of varia-
results are still within acceptable limits, trends in tion in the use of internal QC during the actual ana-
results are key indicators, which can be used to pre- lytical process of FBC testing but did not cover the
vent major nonconformance in test results. This is pre- and postanalytical aspects which should also be
critical when providing quality patient care. part of the total quality control process within a labo-
ratory. The 253 laboratories that participated in this
survey used a wide selection of instrument models
DISCUSSION
and tested from <50 up to >500 patient samples daily.
Internal quality control in a laboratory setting is pri- The majority of laboratories ran a 24-h service or an
marily to ensure the accuracy and precision of test on-call service, which in some cases influenced the

© 2016 John Wiley & Sons Ltd, Int. Jnl. Lab. Hem. 2017, 39, 84–94
92 J. SIOUFI ET AL. | QC PROTOCOL FOR FBC INSTRUMENTATION

quality control process used in their laboratory. This Measurement Procedures: Principles and Definitions;
questionnaire highlighted the variation in the type of Approved Guidelines – Section 8.2)[2], which states:
control material used, how frequently the control ‘Quality control samples must be analysed at least
material was processed and the rules applied to moni- once during each user-defined analytical run length’.
tor any nonconforming event. Manufacturers of analytical systems or reagents may
recommend the number of quality control specimens
and their location within the run. However, manufac-
Part A
turers’ recommendations should be used as guidelines
Most laboratories utilize a range of strategies to ensure and the frequency of QC measurement should be
that performance goals for IQC are met. It is not possi- established by the laboratory considering factors out-
ble to quantify the effectiveness of each platform lined in Section 7.2 (7.2 – Length of Analytical Run)
without a detailed log to capture the detection [2]. Laboratories should take note of these guidelines
method of the QC outlier, such as primary tri-level when establishing the frequency of running control
controls, statistical functions, such as the moving material.
average (XB analysis), or drift control material. Instrument drift implies a change in performance
The majority of laboratories (92.5%) that of the instrument with time. Manufacturers all recom-
responded ran commercial tri-level controls specific to mend the use of secondary control material to moni-
the instrument model on a regular basis to monitor tor instrument drift, suggesting a movement in
the FBC instrumentation. The other 7.5% of laborato- analytical imprecision or bias in between primary QC
ries used generic brands or only use part of the tri- runs. Sixty-five per cent of the laboratories surveyed
level controls, that is they use the normal control run a drift control of various sources and time fre-
only; 47.4% of laboratories that participated in this quencies in conjunction with their tri-level controls.
survey also stated that a drift control was run in con- Fifty-nine per cent of these laboratories stated that
junction with the tri-level controls. the source of the drift control was commercial mate-
The tri-level control material produced by instru- rial and 40% used patient material. The allowable
ment manufacturers is stabilized whole blood which limits for drift controls set by laboratories varied, the
affects the matrix of the cell components to ensure a majority of laboratories using limits set by the manu-
longer shelf life. The principles used by FBC instru- facturer. Laboratories using patient material either
ments to quantify each parameter on stabilized whole based their limits on the commercial control limits for
blood also vary, causing variation in target results a normal sample or established their own range, using
from model to model. This variation should not be the average plus or minus 2 standard deviations. CSLI
evident in patient material when cross-comparative Standard C24-A3 [2] states that the mean and stan-
studies are performed; however, biological variation dard deviation of results for a particular control mate-
may cause very slight shifts in patient results. The rial should be established on the basis of repeated
varying levels of controls are important and should measurements. Control limits can then be calculated
cover the possible range of patient results for all FBC from the means and standard deviation. If commercial
parameters. Laboratories that stated only running con- assayed material is being used, then the values stated
trols of normal levels should be aware that they run on the assay sheet should be used as a guide only and
risks in not controlling FBC parameters in the upper the mean and standard deviation should be estab-
or lower region and may miss nonconforming patient lished by serial testing. It is recommended that labora-
test results. tories refer to these guidelines when setting up
The questionnaire highlighted common practices in allowable ranges for drift controls.
the frequency of running tri-level controls; however, The majority of FBC instruments also provide labo-
this varied according to the number of samples pro- ratories with a built-in control measure, the moving
cessed within the day (Table 1) and how frequently average or XB analysis. This facility monitors the red
the samples were processed. Clinical and Laboratory cell calculated indices, MCV, MCH and MCHC by
Standards Institute have published guidelines (C24-A3 plotting the average of batches of samples, setting
– Statistical Quality Control for Quantitative allowable ranges to monitor drifts or sudden

© 2016 John Wiley & Sons Ltd, Int. Jnl. Lab. Hem. 2017, 39, 84–94
 J. SIOUFI ET AL. | QC PROTOCOL FOR FBC INSTRUMENTATION 93

systematic changes. This facility is a good indicator for where ‘A’ represents the number of control observa-
system failures or problems and will provide the oper- tions and ‘L’ is a control limit derived from Gaussian
ator with alarms to take immediate action if the mov- statistics[2]. For example, the rule 13s will imply that
ing average falls outside the acceptable range. It was action is taken when a single control result is outside
surprising to see that 27.4% of laboratories did not 3 standard deviations from the mean. The rule 22s will
use this functionality, and it is recommended that imply that there are two control results outside 2
they review their internal QC protocol. standard deviations from the mean.
The questionnaire also highlighted the variability
of time frames in which laboratories calibrated their
Part B
instrument. These were subject to the usage of the
instrument; however, the most common practice was The second part of the questionnaire was to review
at six-monthly intervals. It was surprising that one the action taken by laboratories when issues appeared
laboratory stated that they never calibrated their in their EQA survey reports. It was pleasing to see
instrument, and another stated that is was rarely that the majority of laboratories that participated in
required for their Sysmex XS instrument. We strongly this questionnaire indicated the preferable choice or
advise these laboratories to reassess their internal action taken, as recommended by RCPAQAP Haema-
quality practice for recalibration of FBC instruments. tology.
There are requirements for good laboratory practice The majority of laboratories, when asked what
in medical laboratories to sustain quality patient care. action would be taken if results were outside the
With any discipline that derives quantitative test mea- allowable limits of performance, indicated that they
surements, processing IQC samples is essential. It is would review the internal QC on the day of testing
also essential that the internal QC process incorporates their external QA samples. Laboratories that stated
guidelines, so the operator can detect nonconforming they would do repeat testing (24.1%) should be
test results before they are released. aware that these samples have expiry dates and would
The majority of responses returned by laboratories not necessarily yield the correct results, which is not
stated that they used the manufacturer ranges to best practice. The laboratories that stated that they
determine nonconforming test results or they simply would do nothing or recalibrate their instrument
viewed trends. Some laboratories stated that a non- immediately should consider issues that could have
conforming result would be a single result for any caused these outliers before no action or immediate
sample that was outside the manufacturers range. action is taken.
A large number of laboratories used the Westgard This is also the case for questions 2 and 3, with the
rules, which were originally published by Dr James majority of participants taking the expected action.
Westgard in the early 1980s and are based on princi- The importance of participating in PT modules is to
ples of a statistical process. These rules are used by not only to provide peer comparison but also to see
laboratories to view systematic errors by looking at whether trends or sudden shifts in results are reflected
trends or shifts in their IQC sample results. These in your internal QC results.
rules are meant to be warning systems for large sys-
tematic errors so that nonconforming test results can
CONCLUSION
be identified before patient results are issued for
patient care. From the results received from this questionnaire, it is
Rules that are applied to monitor nonconforming evident that there is variation in QC protocol amongst
test results should use Gaussian (normal) distribution laboratories for reviewing the analytical results of IQC
of results and then by calculating the average and and EQA data from FBC instrumentation. The major
standard deviation to determine the control limits, for factor causing this variation was the capacity of the
example the mean plus or minus 3 standard devia- laboratory, that is the number of samples processed
tions. Quality control rules are then applied, such as and whether a laboratory ran a 24-h service. It should
Westgard rules, to monitor systematic errors. These be noted that the appropriate levels of controls that
rules are represented by abbreviation of the form AL, should be processed through a FBC instrument should

© 2016 John Wiley & Sons Ltd, Int. Jnl. Lab. Hem. 2017, 39, 84–94
94 J. SIOUFI ET AL. | QC PROTOCOL FOR FBC INSTRUMENTATION

cover the possible analytical range of the derived FBC this mechanism reviews the mean of the patient pop-
parameters (WCC, RCC, Hb, platelets). How regularly ulation being processed through the instrument, any
IQC samples should be run is subject to certain condi- change in the mean would suggest a change in the
tions as mentioned above; however, laboratories calibration of the instrument, which is a good indica-
should employ a mechanism that will highlight non- tor to investigate possible causes of out-of-control
conforming test results at any time when the FBC issues.
instrument is operational. Laboratories should also be aware of the impor-
A significant finding of this survey was the large tance of participating in an EQA programme and to
number of laboratories using a drift control as well as take appropriate action when either results are outside
tri-level QCs. When reviewing, the cause of drift con- the allowable limits of performance or trends in
trol outliers, analyser or reagent performance was not results are becoming apparent in the cumulative sur-
listed as the most frequent cause (6.6%). It is most vey results. The purpose of EQA is to evaluate labora-
likely to be a result of inappropriate mixing or sam- tory performance against laboratories using the same
pling of the drift material (43%), followed by the QC method (peer group). By participating in EQA pro-
sample integrity (39.5%), suggesting that it con- grammes, you are able to compare results to your
tributes to unnecessary investigations and subsequent peer group and take action when results appear to be
waste of resources. These findings in addition to the out of control, ensuring accuracy of patient test
various approaches used by individual laboratories in results.
terms of what they consider to be sufficient to moni-
tor instrument drift raise questions about the effec-
AC K N OW L E D G E M E N T S
tiveness of running a drift control. This is an area that
warrants further investigation. Members of the RCPAQAP Haematology advisory
It is recommended that laboratories should take committee and to all RCPAQAP participants who took
advantage of the in-built mechanism to control non- part in this questionnaire.
conforming test results on FBC instrumentation. As

REFERENCES Hematology Analyzer; Approved Standard, Principles and Definitions; Approved Guide-
2nd edn. Wayne, PA: CLSI; 2010. line, 3rd edn. Wayne, PA: CLSI; 2006.
1. CLSI H26-A2. Validation, Verification, and 2. CLSI: C24-A3. Statistical Quality Control for
Quality Assurance of Automated Quantitative Measurement Procedures:

© 2016 John Wiley & Sons Ltd, Int. Jnl. Lab. Hem. 2017, 39, 84–94