See discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/7033302
Application of an acceptance sampling plan for post-production quality
control of chemotherapeutic batches in an hospital pharmacy
Article in European journal of pharmaceutics and biopharmaceutics: official journal of Arbeitsgemeinschaft fur Pharmazeutische Verfahrenstechnik e.V · September 2006
DOI: 10.1016/j.ejpb.2006.04.002 · Source: PubMed
CITATIONS
READS
22
2,915
7 authors, including:
Isabelle Borget
Angelo Paci
Institut de Cancérologie Gustave Roussy
Institut de Cancérologie Gustave Roussy
323 PUBLICATIONS 4,914 CITATIONS
241 PUBLICATIONS 3,906 CITATIONS
SEE PROFILE
SEE PROFILE
Stefan Michiels
Lionel Mercier
Institut de Cancérologie Gustave Roussy
Institut de Cancérologie Gustave Roussy
562 PUBLICATIONS 20,374 CITATIONS
26 PUBLICATIONS 344 CITATIONS
SEE PROFILE
Some of the authors of this publication are also working on these related projects:
2-HydroxyGlutarate as surrogate IDH mutations biomarker View project
LACE-Bio2 View project
All content following this page was uploaded by Angelo Paci on 04 December 2017.
The user has requested enhancement of the downloaded file.
SEE PROFILE
Journal of Oncology
Pharmacy Practice
http://opp.sagepub.com/
Safety and quality assurance of chemotherapeutic preparations in a hospital production unit:
Acceptance sampling plan and economic impact
A Paci, I Borget, L Mercier, Y Azar, R.P Desmaris and P Bourget
J Oncol Pharm Pract published online 10 May 2011
DOI: 10.1177/1078155211402865
The online version of this article can be found at:
http://opp.sagepub.com/content/early/2011/05/06/1078155211402865
Published by:
http://www.sagepublications.com
Additional services and information for Journal of Oncology Pharmacy Practice can be found at:
Email Alerts: http://opp.sagepub.com/cgi/alerts
Subscriptions: http://opp.sagepub.com/subscriptions
Reprints: http://www.sagepub.com/journalsReprints.nav
Permissions: http://www.sagepub.com/journalsPermissions.nav
Downloaded from opp.sagepub.com at Institut Gustave Roussy on May 12, 2011
Journal of
Oncology
Pharmacy
Practice
Article
Safety and quality assurance of
chemotherapeutic preparations in a
hospital production unit: Acceptance
sampling plan and economic impact
J Oncol Pharm Practice
0(0) 1–8
! The Author(s) 2011
Reprints and permissions:
sagepub.co.uk/journalsPermissions.nav
DOI: 10.1177/1078155211402865
opp.sagepub.com
A Paci
Pharmacology and Drug Analysis Department and Department of Clinical Pharmacy, Institute
Gustave Roussy, Villejuif, France
I Borget
Service of Health Economy, Biostatistic and Epidemiology Department, Institute Gustave Roussy,
Villejuif, France
L Mercier
Pharmacology and Drug Analysis Department and Department of Clinical Pharmacy, Institute
Gustave Roussy, Villejuif, France
Y Azar
Pharmacology and Drug Analysis Department, Institute Gustave Roussy, Villejuif, France
R.P Desmaris
Department of Clinical Pharmacy, Institute Gustave Roussy, Villejuif, France
P Bourget
Department of Clinical Pharmacy, Institute Gustave Roussy, Villejuif, France
Abstract
Objective.The opportunity to apply a sampling plan was evaluated. Costs were computed by a microcosting study.
Setting. In 2003, a sampling plan was defined to reduce the number of chemotherapy quality controls while preserving
the same level of quality. Recent qualitative and quantitative changes led us to define a second sampling plan supplemented by an economic evaluation to determine the cost and cost-savings of quality control.
Methods. The study considers preparation produced during four semesters classified into three groups. The first one
includes drugs produced below 200 batches a semester. Group 2, those for which the lot of preparation lots would have
been rejected twice among these four semesters. Group 3, those would have been accepted (3 ‘acceptable lot’).
A single sampling plan by attributes was applied to this group with an acceptance quality level of 1.65% and a lot tolerance
percent defective below 5%. A micro-costing study was conducted on quality control, from the sampling to the validation
of the results.
Results. Among 39 cytotoxic drugs, 11 were sampled which enabled to avoid a mean of 17,512 control assays per year.
Each batch of the 28 non-sampled drugs was however analyzed. Costs were estimated at 2.98E and 5.25E for control
assays depending of the analytical method. The savings from the application of the sampling plans was 153,207E in 6 years.
Conclusion. The sampling plan allowed maintaining constancy in number of controls and the level of quality with
significant costsavings, despite a substantial increase in drugs to assay and in the number of preparations produced.
Corresponding author:
Angelo Paci, Pharmacology and Drug Analysis Department, Institute
Gustave Roussy, Villejuif, France
Email: angelo.paci@igr.fr
Downloaded from opp.sagepub.com at Institut Gustave Roussy on May 12, 2011
2
J Oncol Pharm Practice 0(0)
Keywords
Quality assurance, sampling plan, cost, chemotherapy, oncology practice
Introduction
The drug supply chain in health establishments is composed of series of successive stages: drug prescription is
a medical task, preparation and dispensation are pharmaceutical ones, whereas drug administration belongs
to nurses. Potential errors such as wrong dosage, solvent’s errors, wrong labeling or, in some exceptional
cases, chemotherapeutic active substances’ errors can
occur, leading to potential risks of morbidity and mortality for patients. Recommendations and quality
assurance processes have since been applied to ensure
safety at each stage of the drug supply chain.
Concerning pharmaceutical tasks, the pharmacist is
responsible for procurement, distribution, surveillance,
and control of all drugs used in the hospital. Drug preparation and quality control are also mandatory activities for a hospital pharmacy.1 The drug preparation
activity must be performed in conformity with good
practices,2 according to article L. 5121-5 of the Public
Health Code.3 Moreover, drug preparation and quality
control activities performed in a hospital pharmacy
must also adhere to the principles defined in Hospital
Pharmacy Good Practices (HPGP).4 According to
HPGP principles, the final product of drug preparation
must be inspected to verify that it has been accurately
compounded using the correct ingredients and the specific amounts of each ingredient in an appropriate reservoir.5,6 Even if these rules concern all drugs,
particular attention must be paid to fragile or potentially toxic preparations, like sterile compounds, cytotoxic drugs, or preparations for parenteral nutrition.
By allowing the identification of qualitative and
quantitative preparation non-conformity, verification
of the finished product contributes to ensuring drug
supply chain safety. However, these quality controls
were rare in the hospital pharmacy setting. In 2002, a
study showed that only 4.7% of American pharmacies
compounding cytotoxic preparations performed these
controls and were thus compliant with 2000 ASHP
guidelines on quality assurance for pharmacy-prepared
sterile products.7,8
Since 1998, our clinical pharmacy department has
been equipped with an analytical platform using both
high performance thin-layer chromatography (HPTLC)
and high performance liquid chromatography (HPLC)
to ensure post-production quality control of chemotherapeutic preparations, both qualitatively (identification) and quantitatively (concentration).9,10 These
analyzes allow post-production quality control assays
of most of the cytotoxic drug solutions commonly used,
as given in Table 1. For 5 years, more than 75% of the
entire production was analyzed which led to a decline in
the non-conformity (NC) rate from 8.9% to 2.2%. In
2003, faced with the growing number of chemotherapy
preparations to control, the need for systematic quality
control versus sampling of manufactured preparations
arose. The opportunity to reduce the number of control
assays while preserving the same level of quality was
seized by implementing a sampling plan. It used a statistical approach between no control and 100% of controls: the size of the sample to be controlled was
determined for each chemotherapy drug in order to
obtain the same level of quality (expressed as a percent
of NC) in the sample as in all preparations.11,12 The
Table 1. Classification of drugs into three groups
Category of drugs
Drugs concerned
Group 1: drugs produced less than 200 preparations per semester
20 drugs: actinomycin D, asparaginase, carmustine, daunorubincin, dacarbazine, liposomal doxorubicin, fludarabine, idarubicin, melphalan, thioguanine, novantrone, pemetrexed, streptozotocin, thiothepa, paclitaxel,
vindesine, cetuximab, toptecan, mitoxantrone, busulphan
8 drugs: bleomycin, irinotecan, methotrexate, rituximab, docetaxel, vinblastine, vinorelbine, etoposide
11 drugs: epirubicin, 5-fluorouracil, cytarabine, carboplatin, cisplatin, cyclophosphamide, gemcitabine, doxorubicin, ifosfamide, oxaliplatin,
vincristine
Group 2: drugs outside the specification limits for
sampling
Group 3: drugs for which the acceptance sampling plan was applied
Downloaded from opp.sagepub.com at Institut Gustave Roussy on May 12, 2011
Paci et al.
3
possibility of applying the plan was evaluated for nine
drugs whose production exceeded 400 per year. The size
of the sample of the others was too small to obtain
correct estimations of the NC rate. Six chemotherapy
drugs (fluorouracil, cisplatin, cyclophosphamide, ifosfamide, doxorubicin, and epirubicin) were then sampled, allowing a reduction of 50–83% in the sample
size to be controlled. From 2003 to 2007, the sampling
plan enabled us to avoid about 8000 control assays out
of 21,000 per annum, representing 38% of the overall
number of analyzes. Among the nine drugs whose production exceeded 400 per year, three (methotrexate,
cytarabin, and etoposide) were not included in the sampling plan, as they represented a higher risk of NC than
the others: it was decided that all preparations of those
drugs would be controlled.13
However, some major elements in the production
and/or in the control process have changed since
2003. First, the number of preparations produced
annually has increased from 30,000 to 52,000. Second,
39 chemotherapy drugs can now be controlled, resulting both from the availability of new drugs in the
cancer field (trastuzumab, bevacizumab, etc.) and
from analytical procedure development. The number
of preparations of some drugs produced annually has
been significantly modified due to treatment protocol
changes. Some of them are now produced fewer than
400 times per year or conversely. Finally, the overall
NC rate has decreased from 2.2% to 1.65%. The
hypothesis and the estimations that were used for the
application of the first sampling plan were also partially
obsolete. A new sampling plan was necessary, in order
to maintain constancy in the number of control assays
and the level of quality, despite a substantial increase in
the number of therapeutic preparations produced and
cytotoxic drugs to control. Moreover, the unit cost of
post-production quality control has never been precisely estimated.
Thus, the objective of this study was to evaluate the
application of a second sampling plan taking into
account the qualitative and the quantitative changes
in production, as well as the decrease in the rate
of NC. This evaluation was supplemented by an economic evaluation aimed at determining the unit cost of
post-production quality control assessment of chemotherapy preparations in the hospital and at evaluating
the economic impact of the application of the sampling
plans.
Methods
Quality control methods
Our laboratory is equipped since 1998 with an analytical platform composed of HPTLC and HPLC
designed to work in tandem with the production unit.
Chromatographic methods coupled to spectrometric
detection allow undoubtedly the identification of the
drug used in the preparation and concomitantly,
its amount or concentration determination. As previously published, our chromatographic methods
allow the separation of each compound and their identification through the retention time or retention factor combined to their spectra.9 Each method was
validated according ICH guidelines. For HPLC, repeatability and intermediate precision were all below
5% while those for HPTLC were evaluated between
3% and 6%.
Whatever the drug, a preparation was considered
as non-conform (NC) if it contains the wrong drug
(qualitative error), or when its measured concentration was outside the specification limits defined as the
target concentration 10% for two consecutive
assays (quantitative error). If it was not technically
possible to perform a second assay after an initial
non-conform assay and as the real status of
unchecked samples was not known, the preparations
corresponding to these samples were considered as
non-conform.
Application of the second sampling plan
The possibility to apply the new version of the acceptance sampling plan was evaluated for the 39 chemotherapy drugs whose analytical assay was available,
based on data accumulated over four semesters (years
2005 and 2006).
The statistical analysis took into account ‘preparation lots’ composed by all the preparations (n) produced for each drug during one semester. Cytotoxic
drugs were then classified into three groups. The first
group was composed of drugs whose production was
below 200 preparations per semester. The size of the
‘preparation lot’ of these drugs (group composed of
drugs produced 17–188 times per semester) was too
small to obtain correct estimations of the NC rate.
No sampling plan was applied to this group, meaning
that each preparation was controlled. The possibility
of applying the sampling plan was then evaluated for
drug preparations exceeding 200 preparations per
semester, by applying acceptance sampling by attributes. It consisted of counting the total number of NC
preparations (d) in the ‘preparation lot’ of size (n). If
(d) was not greater than the acceptable number of
NC defined by the plan (c), the ‘preparation lot’
was evaluated as an ‘acceptable lot’; otherwise it
was marked as an ‘unacceptable lot’. Thus, the
number of acceptable lots for each drug ranged
from 0 (all the lots corresponding to the four semesters were unacceptable) to 4 (all the lots
Downloaded from opp.sagepub.com at Institut Gustave Roussy on May 12, 2011
4
J Oncol Pharm Practice 0(0)
corresponding to the four semesters were acceptable).
Group 2 was composed of drugs for which the ‘preparation lot’ would have been rejected (NC rate higher
than 1.65%) at least over two semesters. It was
decided that all preparations of this group of drugs
would be controlled instead of applying a sampling
plan, in order to obtain a correct estimate of the NC
rate.
Group 3 was composed of drugs whose ‘preparation
lot’ would have been accepted at least three out of four
semesters (3 ‘acceptable lot’). The sampling plan was
only applied to this group. The methodology which
allowed us to determine the optimal sample size to control was similar to that used to apply the first sampling
plan.13 The AQL, represented the maximum percentage
of NC that the pharmacist could tolerate to consider
the process as satisfactory. It was fixed at 1.65%, as it
represented the maximum percentage of non-conform
preparations observed over the period 2005–2006. The
pharmacist would like to design a sampling plan such
that there is a high probability of accepting a preparation lot that has a NC rate less than or equal to the
AQL.13
The lot tolerance percent defective (LTPD), i.e. the
risk of accepting a ‘preparation lot’ of unacceptable
quality was fixed at 5%. The acceptance sampling
plan was then applied as follows: based on the
number of preparations (ni) prepared during the last
semester (second semester 2006), (ni) was decreased by
increments of 500 or less in order to calculate the
number of preparations requiring a control assay,
while maintaining the AQL at the targeted value
(1.65%) and an LTPD of less than 5%.
Cost determination
Costs were computed from the hospital viewpoint
(Gustave Roussy Institute) and were expressed in
Euros in 2007. A micro-costing study, restricted to
the resources consumed, was conducted to perform a
control, from the sampling when the chemotherapeutic
preparation was manufactured until the validation of
the results by the pharmacist and the transmission of
the results. Costs attributed to the production of the
preparation (cytotoxic drugs, personnel costs for the
production etc.) were not included. The total cost
of post-production control assays at IGR corresponded
to the sum of direct and indirect medical costs.
Direct medical costs corresponded to materials,
person costs, and the analytical equipment,
whereas indirect costs included overheads and logistics
costs.
The type and the number of medical devices, consumables, material, and reagents consumed annually
were collected retrospectively over three consecutive
years (2005, 2006, and 2007) and were expressed per
control assay. The unit prices of devices according to
the 2007 price list at IGR were used for their valuation.
Personnel costs were calculated using the mean lengths
of time spent performing the different stages of a
series of quality control assays, which were estimated
prospectively for 83 series of control assays performed over 14 days in March 2008. For each
stage, two measurements of the length of time were
estimated: the first one estimated the total length of
time spent performing the given stage (including the
time spent by the technician and the automatic processing time). The valuation of personnel costs was
based on the second estimation, which only considered the length of time needed for the technician to
perform the given stage (the automatic processing
time spent at the given stage was not valorized,
that is, a value was not assigned to it). The valuation
of personnel costs used the hourly employee wage by
professional category. The laboratory technician was
involved in each stage of the post-production control
process, except for the validation of the results which
was done by the pharmacist.
The analytical platform consisted of the CAMAGÕ
station and the HPTLC platform, both purchased in
1998. The cost of the analytical equipment per control
assay and per method was calculated by taking into
account their respective acquisition cost, their depreciation charges (amortization period of 10 years), and an
estimated number of control assays performed until the
end of that period. Indirect costs included overheads
and logistics costs. According to the IGR cost accounting system, indirect costs represented 33% of the
amount of direct costs.
The number of control assays avoided, since the
application of the first sampling plan in 2003 was determined. For each year, it corresponded to the difference
between the number of preparations produced and the
number of control assays performed annually. The
economy generated was then estimated by multiplying
the total number of control checks avoided, since 2003
by the unit cost per control assay.
Results
Application of the second sampling plan
In 2006, the analytical platform was able to assess qualitative and quantitative analyzes for 39 cytotoxic drugs
(30 by HPTLC and 9 by HPLC). Fewer than 200 preparations of 20 of these 39 cytotoxic drugs were produced per semester (representing 3093 out of 31,890
(9.7%) preparations produced in 2006); hence, they
were classified in the first group based on the small
Downloaded from opp.sagepub.com at Institut Gustave Roussy on May 12, 2011
Paci et al.
5
Table 2. Determination of the optimal sample size by applying the acceptance sampling plan to the 11 drugs in the third group
Epirubicin
5-Fluorouracil
Cytarabine
Carboplatin
Cisplatin
Cyclophosphamide
Gemcitabine
Doxorubicin
Ifosfamide
Oxaliplatin
Vincristine
Total
Preparations
produced per
year (n)
Preparations
analyzed per
year (n)
Rate of reduction
of preparation
analysis (%)
Controls avoided
by the plan
per year (n)
832
4813
1658
1260
3367
1512
614
2098
3315
527
1347
21,343
200
630
355
217
414
253
244
382
367
259
510
3831
75
85
67
75
88
83
50
80
88
50
50
632
4183
1303
1043
2953
1259
370
1716
2948
268
837
17,512
‘preparation lot’. No sampling was performed on this
group (Table 1). Eight cytotoxic drugs were classified in
the second group, as the ‘preparation lot’ would have
been rejected over at least two semesters. The sampling
plan was not applied to this group of drugs. This means
that each of the 7454 preparations produced had to be
controlled (23.3% of overall production). The sampling
plan was applied to 11 cytotoxic drugs which constituted the third group whose production attained 21,343
preparations (66.9% of overall production) in 2006.
The application of the sampling plan showed that it
was possible to reduce preparation analysis by
50–88% for these 11 drugs (Table 2). In 2006, as the
estimated number of quality control assays was 3831,
the application of the sampling plan to these drugs
made it possible to avoid 17,512 control assays (55%
of overall production).
Unit cost of quality control checks
From 2004 to 2006, the mean annual expenditures for
material, consumables, and reagents was estimated at
22,794 3301E, for a mean number of 24,929 2507E
quality control assays performed annually. The mean
cost of the resources consumed for performing a control assay was then estimated at 0.91 0.08E. The
mean lengths of time spent performing a control
assay by stage and analytical methods are presented
in Table 3. The total lengths of time (including both
the technician and automatic processing times) were
estimated at 3.6 1.2 and 16.0 5.4 min for assays performed by HTPLC and by HPLC, respectively. During
this time, the technician was occupied, respectively, for
2.1 0.7 and 2.9 2.2 min performing an HPTLC and
an HPLC control assay. The personnel costs for performing a post-production assay were 1.00 0.19E and
1.35 0.92E for HPTLC and HPLC control assays,
respectively. Concerning the cost of equipment, the
acquisition cost of the CAMAGÕ station amounted
to 63,093E. Based on a mean annual number of
18,332 HPTLC control assays per year and a damping
period of 10 years, the cost attributed to the acquisition
and the damping of the analytical equipment for performing an HPTLC control assay was then estimated at
0.33E per control assay. The acquisition cost of the
HPLC platform was 20,747E. With an average of
1228 control assays performed annually and a damping
period of 10 years, the mean cost of acquisition and
damping of the HPLC platform was calculated at
1.69E per control assay. The mean direct costs of control assays were, respectively, 2.24 0.19E and
3.95 0.92E for HTPLC and HPLC. By including
the indirect costs, the total cost for performing a
post-production control assay amounted to, respectively, 2.98 0.25E and 5.25 1.22E (Table 4).
Economic impact of applying the sampling plans
Since 2003, 201,787 preparations have been produced.
As the analytical process was not yet available, 5692 of
these preparations were not controlled. During this
period, the total number of preparations produced
and for which a control was available was then equal
to 196,095. Among this, 146,927 quality control assays
were performed from 2003 to 2008, resulting in a total
cost for the hospital estimated at 457,824E. The application of the sampling plan then enabled the avoidance
of 49,168 control assays (representing 25.1% of the
Downloaded from opp.sagepub.com at Institut Gustave Roussy on May 12, 2011
6
J Oncol Pharm Practice 0(0)
Table 3. Mean lengths of time spent performing an assay by stage and by method (minutes)
HPLC
Separation of dosage and storage (if necessary)
Preparation of the analysis plan
Dilution
Deposit
Chromatography
Revelation
Analysis
Interpretation
Printing
Transmission of results
Preparation of laboratory (preparation of reagents,
maintenance, waste, control tube)
Total
Total time
(technician + automatic
process)
Technician’s
time alone
Total time
(technician +
automatic process)
Technician’s
time alone
0.18 0.30
0.34 0.19
1.06 0.96
–
13.38 3.38
–
–
0.51 0.13
0.41 0.24
0.06 0.05
0.05 0.11
0.18 0.30
0.34 0.20
1.06 0.96
–
0.25 0.25
–
–
0.51 0.13
0.41 0.24
0.06 0.05
0.05 0.11
0.12 0.06
0.37 0.07
0.70 0.12
0.91 0.21
0.34 0.17
0.18 0.15
0.26 0.07
0.31 0.08
0.34 0.16
0.04 0.04
0.06 0.09
0.12 0.06
0.37 0.07
0.70 0.12
0.05 0.01
0.05 0.01
0.05 0.01
0.05 0.01
0.31 0.08
0.34 0.16
0.04 0.04
0.06 0.09
15.99 5.36
2.86 2.24
3.62 1.22
2.12 0.68
Table 4. Cost for performing an HPLTC and an HPLC control
assay (euros)
Direct medical cost (E)
Consumables, reagents, and material cost
Personnel cost
Equipment cost
Total direct cost (E)
Indirect costs (E)
Total cost (E)
HPTLC
HPTLC
HPLC
0.91
1.00
0.33
2.24
0.74
2.98
0.91
1.35
1.69
3.95
1.30
5.25
HPTLC, high performance thin-layer chromatography; HPLC, high
performance liquid chromatography.
production that can be assayed). About 24,841 control
assays were avoided from 2003 to 2006 (first plan),
whereas the application of the second plan in 2006
allowed the hospital to avoid 24,327 control assays in
2 years (Figure 1). The savings generated by the application of the sampling plans were estimated at
153,207E in 6 years.
Discussion
Post-production quality control of chemotherapy is not
a common practice in hospital pharmacy units. In our
institution, it has been developed since 1998, to contribute to an overall quality assurance program started in
2000 at the IGR and is designed to ensure the safety of
the preparation of anti-neoplastic agents and of their
administration conditions. During the first 4 years, routine assays of manufactured preparations reached considerable maturity with 28 cytotoxic drugs controlled
and 23,000 assays performed per year. In 2003, faced
with an increase in the number of preparations produced, the acceptance sampling plan was decided as
an analytical strategy to maintain constancy in the
number of control assays and the level of quality. It
was applied to six cytotoxic drugs, and resulted in the
avoidance of about 8000 control assays of the 26,000
preparations produced annually. First, the decrease in
the number of control assays allowed us to save time,
which was used to develop analytical control assays of
drugs recently introduced in cancer treatment and/or
which had not yet been assayed. From 2003 to 2006,
the number of drugs available for assay thus increased
from 23 to 39. The analytical platform had therefore
reached a high level of exhaustivity, as 98% of overall
production could be qualitatively and quantitatively
assayed. Second, it allowed us to focus our attention
on the other drugs that were systematically controlled.
Corrective measures were implemented, resulting in an
improvement of overall quality (a decrease in the rate
of NC from 2.2% to 1.65%) despite a continuous
increase in the number of preparations produced
(+1400 preparations produced annually). The corrective actions that have been introduced were a review of
the fabrication processes’ and a vocational training of
the personnel. The results of the dosage were used as
indicators of quality that involve salaries in an
approach of continuous improvement. However, the
increase in the number of drugs controlled also led to
Downloaded from opp.sagepub.com at Institut Gustave Roussy on May 12, 2011
Paci et al.
7
50000
Controls not yet available
Controls avoided by the
sampling plan
Controls performed
40000
30000
20000
2nd SP
10000
0
2000
1st SP
2001
2002
2003
2004
2005
2006
2007
2008
Years
Figure 1. Evolution of the number of preparations produced and controlled from 2000 to 2008.
an increase in the number of control assays (a series of
23 control assays per quality control check). At the
same time, HPGP recommended that the production
of cytotoxic preparation henceforth be totally centralized in the Pharmacy Department of the hospital, this
organization being mandatory for the reimbursement
of expensive cytotoxic drugs. This centralization at
the IGR translated into a considerable increase in the
number of preparations produced (from 31,910 to
51,643 from 2006 to 2008). By responding to both qualitative and quantitative changes in production, the
second sampling plan allowed us to maintain constancy
in the number of control assays performed, since the
application of the first plan, by taking into account the
new rate of NC. As the number of drugs produced
more than 200 times per semester had increased, 11
cytotoxic drugs were now concerned by the sampling
plan and the number of control assays avoided was
17,000 per year. Without implementing the second sampling plan, the Centralized Pharmacy Department
would have had to have set up alternative solutions
to deal with the increase in the number of control
assays. The recruitment of a second technician and/or
the acquisition of a second analytical platform could
have been considered but those are expensive alternatives. Some hospital pharmacies have chosen to control
only some cytotoxic drugs, while other drugs are not
controlled. This provides only a partial estimation of
the rate of NC and criteria used to determine the selection of drugs to control are unclear (based on the
number of preparations produced or analytical
method development).
In our institute, preference was given to the efficiency
of the analytical platform in terms of a constant level of
human and material resources. Based on our estimation, the sampling plan will allow us to avoid between
30,000 and 50,000 control assays among the 80 to
90,000 chemotherapy preparations that will be produced annually in the very near future. The sampling
plan has also made it possible to continue the routine
inspection of every preparation of drugs that is not
included in the sampling plan, and thus to continue
to improve their quality and reduce the total rate of
NC. This systematic inspection of drugs which are
not included in the plan (groups 1 and 2) is also possible as they represented only 33% of the total preparations produced. The quality assurance process engaged
by the Department also includes estimating quality
indicators, providing results in bi-annual meetings
and reflecting upon the manufacturing process with
the pharmaceutical and analytical teams.
This article also aimed at estimating the unit cost of
post-production quality control for the hospital.
It showed that the costs of quality control by HPTLC
and HPLC were, respectively, estimated at 2.98E and
5.25E and appeared to be similar to that of other
Downloaded from opp.sagepub.com at Institut Gustave Roussy on May 12, 2011
8
J Oncol Pharm Practice 0(0)
biological determinations performed in the hospital.
Cost drivers were the use of personnel and consumables. Depreciation charges were low in our case
because the number of control assays was substantial.
HPLC determinations were more expensive than those
performed by HPTLC but they only slightly affected
the total cost of post-production quality control, as
HPLC determinations represented only 6% of the
assays performed. Some authors described recently
the use of rapid HPLC method which could reduce
the cost.14 However, the identification was based only
on the UV spectrum of the drug as the time of analysis
was to short for a real separation. As UV detection is
not considered as an efficient analytical method alone
for identification of an active substance, we preferred
the use of a less rapid HPLC method to separate analytes. To circumvent the lack of selectivity of single
UV-detection, we are currently using a more rapid
and accurate analytical method, which uses infra-red
detection combined to a UV one (IRTF–UV). This
new technology reduces considerably the time of analysis without any doubt of identification for most of the
drugs used in onco-hematology. We plan to investigate
its economic impact on our quality assurance program
in the next future.
Excepted the indirect costs, which were specific to
our institution, the unit costs used for the valuation
corresponded to those of the market. Our results are
therefore generalizable to other hospitals and allowed
us to inform pharmacists and hospital managers about
the economic impact of post-production quality control. However, even if these are mandatory activities,
the pharmaceutical activities of drug preparation or
quality control are not coded like other medical activities (e.g., radiology or biology). Thus, the hospital does
not receive any financial compensation when it conducts these control checks: control check costs represent exclusively an item of expenditure for a hospital.
The sampling plan contributed to hospital efficiency, as
it both allowed us to maintain the same level of quality
and to reduce the overall cost of quality control. Based
on our estimation, the application of the sampling plan
enabled savings of about 150,000E over the past 6
years, representing 25% of the overall budget of this
activity. Finally, control check costs appeared to be
very low as compared to potential litigation costs,
which can occur in case of errors during the production
of chemotherapy preparations.
References
1. Decree n 2000–1316, 2000.
2. Projet de Guide de Bonnes Pratiques de Préparations.
Agence Française de Sécurité Sanitaire des Produits de
Santé (A.F.S.S.A.P.S.), Bonnes Pratiques de Préparation,
Paris, 21st November, 2007.
3. Article L. 5121–5 of the Public Health Code.
4. Arrêté du 22 juin 2001 et B.O. MES n 2001-BOS 2 bis,
2001.
5. Boylan JC. Essential elements of quality control. Am J
Health-Sys Pharm 1983; 23: 1936–1939.
6. American Society of Health System Pharmacists. ASHP
guidelines on preventing medication errors with antineoplastic agents. Am J Health-Syst Pharm 2002; 29:
1648–1668.
7. American Society of Health System Pharmacists. ASHP
guidelines on quality assurance for pharmacy-prepared
sterile products. Am J Health-Syst Pharm 2000; 57:
1150–1169.
8. Morris AM, Schneider PJ, Pedersen CA and Mirtallo
JM. National survey of quality assurance activities for
pharmacy-compounded sterile preparations. Am J
Health-Syst Pharm 2003; 60: 2567–2576.
9. Bourget P, Paci A, Rey JB, Mercier L and Dermidjian
S. Contribution of high-performance thin-layer chromatography to a pharmaceutical quality assurance program
in a hospital chemotherapy manufacturing unit. Eur J
Pharm Biopharm 2003; 56: 445–451.
10. Gravel E, Bourget P, Mercier L and Paci A.
Fluorescence detection combined with either HPLC or
HPTLC for pharmaceutical quality control in a hospital
chemotherapy production unit: application to camptothecin derivatives. J Pharm Biomed Anal 2005; 39:
581–586.
11. Weber RT. An easy approach to acceptance sampling:
how to use MIL-STD-105E, 1st ed. Madison, WI:
American Society for Quality Press, 1991.
12. Montgomery DC. Introduction to statistical quality control, 4th ed. New York: Wiley, 2000.
13. Borget I, Laville I, Paci A, Michiels S, Mercier L,
Desmaris RP, et al. Application of an acceptance sampling plan for post-production quality control of chemotherapeutic batches in a hospital pharmacy. Eur J Pharm
Biopharm 2006; 64: 92–98.
14. Delmas A, Gordien JB, Bernadou JM, Roudaut M,
Gresser A, Malki L, et al. Quantitative and qualitative
control of cytotoxic preparations by HPLC-UV in a centralized parenteral preparations unit. J Pharm Biomed
Anal 2009; 09: 1213–1220.
Downloaded from opp.sagepub.com at Institut Gustave Roussy on May 12, 2011
View publication stats