Content uploaded by Daniel Gartner
Author content
All content in this area was uploaded by Daniel Gartner on Oct 03, 2020
Content may be subject to copyright.
Modelling of the Collections Process in the Blood Supply
Chain: A Literature Review
Emily P. Williams1, Paul R. Harper1, and Daniel Gartner1
1School of Mathematics, Cardiff University, Senghenydd Road, Cardiff
Initial Submission: April 9, 2019; 1st Revision: October 20, 2019; 2nd Revision: March 30, 2020
Accepted: May 22, 2020
Abstract
Human blood is a scarce resource and its role in healthcare is fundamental, with donated
blood saving the lives of many on a daily basis. The blood supply chain is responsible for the
transfer of blood from donor to the recipient, but the availability of such an invaluable resource as
human blood is ultimately attributable to the many voluntary donors. Thus, the efficiency of the
collection of donated blood is crucial to the downstream effectiveness of the blood supply chain.
We provide a detailed review on the use of quantitative methods for the process of blood collection
from donors. We describe the functional areas which are appointment scheduling, collection
policy, crisis situation, donor demographics, location/clinic planning, staff utilisation and vehicle
routing. Furthermore, we analyse the existing literature with regards to methods, modelling
objectives and the planning levels such as strategic, tactical and operational. Finally, we break
down the articles into whether or not case studies lead to the implementation of the methods in
practice. In total, we review 46 relevant publications on the intersection between OR/MS and
other disciplines. We use our presented framework to categorise the existing approaches and
highlight gaps such as scheduling of both staff and appointments for blood donation clinics.
Keywords: blood supply chain, blood collection, operational research, healthcare modelling,
literature review
1
1 Introduction
Blood is a scarce resource and its role in healthcare is fundamental, with donated blood saving the lives
of many on a daily basis and on a global scale. It is estimated that 85 million blood transfusions are
carried out annually across the globe, which translates to nearly 3 blood transfusions per second [15].
The blood supply chain is responsible for the transfer of blood from donor to the recipient, but the
availability of such an invaluable resource as human blood is ultimately attributable to the many
voluntary donors. Blood is a complex product due to its perishability and compatibility requirements
between donor and patient. The blood supply chain comprises of four echelons: (Osorio et al. [37])
Collection: This is the beginning of the blood supply chain and involves either fixed or mobile
blood donation clinics, or a combination of both. Blood products may be collected in the form
of either an individual blood component via a process called apheresis, or whole blood which is
the most common form of donation. From a donor’s perspective, a clinic typically follows the
process displayed in Figure 1, with queues often arising between the activities (which will be
referred to in this literature review as workstations). Eligibility screening is necessary as there
are many restrictions on who may donate blood [47]. Screening often becomes a prolonged
process with many health-related questions and a test for iron levels within the donor’s blood.
The collection echelon ends with all blood donation units transported to a blood processing
centre.
Donor Arrives
at Clinic
Registration Eligibility
Screening
Donation of
Blood Refreshment Donor Leaves
Clinic
Figure 1: Typical Path of a Donor at a Donation Clinic
Production: This takes place at a blood processing centre, where donated units of blood
are tested and separated into various components (red blood cells, platelets and plasma) as
required. The production of platelets depends on the amount of time since the donation, as
platelets must be separated from whole blood shortly after collection. This echelon ends with
blood being packaged ready for distribution and moved to storage.
2
Inventory: Storage of blood products may either take place at a blood processing centre or
a stock-holding unit. Each type of blood product has unique shelf-lives and specific storage
requirements, with platelets being the most complicated; platelets must be kept in an agitated
state at ambient temperature for a maximum of ≈5 days.
Distribution: This echelon consists of the preparation of orders of blood products and the
transportation of such orders to the respective hospitals. Decisions involved include the dates
and blood types of dispatched products, due to blood compatibility and possible limited inven-
tory.
Currently, the vast majority of literature concerns the inventory stage (Osorio et al. [37]) with little
in-depth research carried out on the collections process. Therefore, the aim of this review is to evaluate
the existing literature that deals with modelling of blood collection, provide a detailed classification
of the selected articles, and thus identify any areas that may benefit from further research.
The remainder of this literature review is structured as follows. In Section 2, we describe how we
conducted the structured search and provide an overview of previous literature reviews. In Section 3,
we describe the relevant characteristics of blood collections. We demonstrate how the retrieved articles
from Section 2 were classified into the various categories. Section 4 closes our literature review with
conclusions.
2 Selection Criteria and Previous Reviews
2.1 Selection Criteria and Search for Relevant Literature
We search for journal publications from the Clarivate Analytics Journal Citation Report (JCR) in
the subject categories of Health Policy and Services (HPS), Medical Informatics (MI), Industrial
Engineering (IE), as well as Operations Research and the Management Sciences (OR/MS). These
categories are selected due to the complex nature of the blood supply chain, with modelling of such
arising from various fields and disciplines. The rationale for choosing these categories is to capture
literature from a range of perspectives, whether it be policy decisions and service improvement (HPS),
3
information systems and data mining (MI) or quantitative healthcare engineering (IE). The inclusion
of these categories, together with OR/MS, yields a thorough analysis of research contributing to
improvement of the collection echelon of the blood supply chain.
Using a structured search string, Scopus provided a base set of articles with the search mainly
focusing on “blood collection” and “blood donation”. The search excludes all publications before 1996
and produced a total of 106 results in February 2019. All articles that did not specifically refer to the
collections process were considered irrelevant, as were all articles that did not contain quantitative
analysis or discuss optimisation of the blood supply chain. A forwards and backwards search from
relevant articles was then conducted (as suggested by Webster & Watson [50]) and all relevant results
collated. In carrying out our forwards search, we not only retrieved journal publications citing the
articles from the original search but also a PhD thesis (van Brummelen [13]) which we decided to
include in our final (relevant) set of publications. Collectively, this gives a total of 46 articles, as
shown in Figure 2.
Or iginal Search
106
Relevant Findings
28
Forwar ds Search
208
Backwar ds Search
766
Relevant Findings
12
Relevant Findings
6
Tot al Rel evant
Fi ndi ngs
46
Figure 2: Scopus Search Results
2.2 Previous Reviews
There is a vast amount of literature that addresses modelling of the blood supply chain, however,
only a minority of articles focus on the collection echelon. From our findings, there are four relevant
reviews that address modelling of blood collection from donors; [6,8,9,37]. The most specific of these
is written by Ba¸s G¨ure et al. [6] who state that there are still several aspects of blood collection
4
that have not yet been explored from an optimisation perspective, but specifically focus on donation
appointment scheduling.
Ba¸s G¨ure et al. also published another review [8], prior to the above article, which studies existing
literature surrounding the blood supply chain in general, categorising by echelon and perspective of
research. Similarly, Osorio et al. [37] provide a review of the whole blood supply chain, with a brief
insight into the literature considering the collections process, categorisation of its planning decisions,
and quantitative models of the process. In both reviews, the section addressing collections however
is significantly smaller than those regarding other echelons of the supply chain, indicating that there
is less existing research covering this particular echelon.
Lastly, Beli¨en and Forc´e [9] deliver a review of the whole blood supply chain also, but instead
categorise by methods used, blood product considered, etc. rather than by echelon of the supply
chain. This makes it difficult to identify areas within sections of the blood supply chain that require
further research.
Our review differs from these previous reviews as it provides a detailed taxonomy of articles, specif-
ically focussing on the collection echelon of the blood supply chain. It includes all existing research on
blood collection that consider an OR approach, including those from an interdisciplinary perspective,
and provides an up-to-date analysis of surrounding literature due to the significant increase in pub-
lications within this field over recent years, as demonstrated by Table 1. Thus, our literature review
presents an in-depth discussion of current literature on the collections process and clearly identifies
areas that require further research.
3 Classification of Literature
Employing the selection criteria, a total number of 46 articles were retrieved and considered relevant
for this review. The articles can be categorized by publication year as given in Table 1. The table
reveals that this field of research is becoming increasingly popular, since 76% of the relevant articles
were published within the last five years. Earlier publications mainly consist of more simplistic
statistical analyses, such as the impact of sharing blood donor deferral registries to streamline donation
5
clinics [20]. However, in 1996 Jacobs et al. [29] published an integer programming method to advise
the American Red Cross on whether to relocate a permanent facility. In more recent years, integer
programming methods remain popular in optimising blood collection, in addition to many other OR
methods including simulation and machine learning. The dramatic increase in research in optimisation
of blood collection may be attributed to the technological advancements which enable the growing
range of OR methods to be utilised effectively.
Table 1: Number of Articles by Publication Year
Time Period Before
2005
2005–
2007
2008–
2010
2011–
2013
2014–
2016
2017–
2019
Total
Number of Articles 2 2 2 5 14 21 46
The articles can also be classified by geographic location, as seen in Table 2. Here, articles are
classified under either the location of any mentioned case studies, or that of the first named author.
The majority of research in blood collections has been based in Asia, and most of these are in
developing and emerging nations (such as Iran) and consider the blood collections process in the
event of disasters (from natural or man-made causes) e.g. earthquakes. However, there is a lack of
research regarding the blood supply chain in the event of disasters from elsewhere in the world, and
this suggests that there is further research to be done to aid the blood services in countries who may
face similar circumstances. There are a significant amount of articles concerning modelling of blood
collection from Europe and America (North and South America collectively) which indicates that this
is indeed a worldwide issue.
Table 2: Number of Articles by Continent
Continent America Asia Europe Rest of
World
Total
Number of Articles 9 24 12 1 46
The articles may also be classified according to their respective JCR (Journal Citation Report)
category to illustrate the approach and perspective of the research. The four subject categories
are Health Policy and Services (HPS), Industrial Engineering (IE), Medical Informatics (MI) and
Operations Research and Management Sciences (OR/MS). Only a total of 29 articles are listed in
6
Table 3, with the remaining 16 articles from other journals outside of these categories (all of which are
from the forwards and backwards search) reinstating the multi-disciplinary nature of blood collection.
The OR/MS category is clearly the most popular of the four categories; this conveys the usefulness
of OR methods to tackle modelling of blood collection from donors.
Table 3: Articles by JCR Category
JCR Category Articles Total
HPS [3, 20, 40] 3
IE [6, 21, 34, 36, 37, 41, 44] 7
MI [19, 49] 2
OR/MS [5, 7, 9, 10, 14, 18, 22, 24, 26, 29, 30, 33, 38, 43, 45, 46, 51] 17
In the next subsections, a classification framework for these articles will be provided, categorising
into functional areas, methods and approaches, planning decision level, and whether a case study was
incorporated.
3.1 Functional Areas Considered
We now classify the articles by the functional area in which the research aims to improve/target. The
functional areas have been organised into the following categories:
Appointment Scheduling: All articles that discuss an appointment scheduling policy, frame-
work, or optimisation of appointments.
Collection Policy: This covers all research and analysis of the way in which blood is collected
from donors, including eligibility/deferral of donors and also the collection strategy.
Crisis Situation: This involves all research into the blood supply chain from the perspective of
a crisis occurring and emergency aid being required e.g. natural disasters such as earthquakes.
Donor Demographics: This includes all analysis and research of donor demographics such as
donor behaviour, location, age, blood type, etc.
7
Location/Clinic Planning: This category includes all research which considers the location
of either temporary or permanent facilities used for blood collection (mostly location of donation
clinics). Both the allocation of clinics and relocation of facilities are categorised under this.
Staff Utilisation: This category includes all articles which consider the allocation of staff to
clinics and also analysis of staff level requirements and skill mix.
Vehicle Routing: All articles that consider routing of vehicles that transport blood and re-
sources for blood collection.
Note that articles which do not fit into any of the above categories are literature reviews and thus
not proposing any particular methods or action.
Table 4: Functional Areas of Research
Functional Area Articles
Appointment Scheduling [3, 7, 13, 36]
Collection Policy [3, 5, 19, 20, 23, 26, 33–35, 38, 40]
Crisis Situation [18, 21–23, 25, 30, 32, 46, 48]
Donor Demographics [2, 3, 19, 22, 31, 43, 49]
Location/Clinic Planning [2, 4, 5, 14, 16–18, 21–27, 29, 30, 34, 39, 42–46, 48, 51, 52]
Staff Utilisation [2, 3, 11, 13, 49]
Vehicle Routing [24, 32, 36, 41, 42, 45]
3.1.1 Appointment Scheduling
Appointment Scheduling is the least popular category, as revealed by Table 4. While many donation
sites worldwide accommodate for unbooked or “walk-in” donors, appointment systems provide an
opportunity to control the arrival patterns of donors and better manage resources and inventory.
Mobasher et al. [36] consider how many donations should be collected within certain time intervals
in accordance with vehicle routing to maximise donations viable for platelet production. Alfonso et
al. [3] place more of an emphasis on appointment strategies regarding frequency of apheresis donations
throughout the day, whilst Ba¸s G¨ure et al. [7] focus on pre-allocating appointment slots to each blood
type at a permanent facility. Van Brummelen [13] considers the combination of appointments and
8
“walk-ins” and produces an optimal appointment schedule based on minimising waiting times for
donors.
More generally, the literature surrounding appointment scheduling in healthcare neglects mainly
strategic decisions but also most tactical decisions; according to Ahmadi-Javid et al. [1] the vast
majority of publications concerning optimisation of outpatient appointments focus on an operational
level. Contrastingly, the publications focusing on blood collection take a more tactical approach.
Ahmadi-Javid et al. [1] also discuss the increasing interest in appointment scheduling, as over 73% of
the publications considered in their review were produced between 2012 and 2016 (the most recent
at the time of publication). Despite this, there is still a significant lack of research applied to blood
donor appointment scheduling.
There is also a lack of research that considers combining various aspects of appointment scheduling.
Reducing waiting times, matching supply to demand (including blood type specific demand) and
managing donor arrivals are all key elements of effective and efficient blood collection, yet research
within appointment scheduling often focus on just one of these goals.
3.1.2 Collection Policy
Collection policy is considerably widely researched, with a significant amount of these articles con-
sidering how many units of blood should be collected at a clinic [5, 23, 26, 33–35, 38, 40]. This is
considered from a variety of perspectives, such as the number of mobile clinics to deploy [26], how
much to collect at each clinic under various scenarios [5], or when to stop collecting each day [35].
Lowalekar & Ravi [34] use the Theory of Constraints (TOC) thinking process to evaluate the collec-
tions process associated with a blood bank in Chennai, India, to identify areas for improvement in the
collection policy and thus improve inventory management. Donor deferrals are observed [3,19, 20, 34],
with Custer et al. [19] evaluating blood safety and policy decisions to assess the impact of deferrals
on inventory. Various methods of blood product collection are also studied [3, 38, 40], with Osorio et
al. [38] optimising the balance between whole blood and apheresis, considering cost and the number
of donors required to reach demand.
9
3.1.3 Crisis Situation
Of the nine articles that address the blood supply chain in the event of a crisis situation, the majority of
these also include location/clinic planning [18,21–23,25,30,46, 48] with the remaining article focussing
on vehicle routing [32]. This is due to the importance of facilities (collection and processing) being
located in an accessible area for a responsive and reliable supply chain. The aims of these articles vary
from minimising transport time [21,22] to ensuring fairness in the distribution of blood products [18].
Most of the publications include a case study [18, 22, 23, 25, 30, 46] with all of these based on the
scenario of an earthquake in various cities in Iran. There is a lack of research within this area from
elsewhere in the world, particularly studying disasters with anthropogenic causes.
3.1.4 Donor Demographics
Donor demographics and behaviour have a significant impact on the success of the collections echelon,
and while many articles indirectly consider these aspects, only seven directly incorporate such aspects
into their models. Alfonso et al. consider donor behaviour such as generosity and availability to
inform their location-allocation model [2] and simulation model [3]. Custer et al. [31] study donor
demographics regarding likelihood of donor deferrals, while Testik et al. [49] study donor arrival
patterns. The location of donors is considered in various location planning models [2, 22, 43] to inform
where to locate clinics to ensure the required amount of blood will be collected. Lastly, Lee &
Cheng [31] classify disparities in donor behaviour to identify possible causes of decreasing donations
and predict donors’ intentions.
3.1.5 Location/Clinic Planning
Location and clinic planning is the most popular category within the selected literature. This is
unsurprising since the success of a clinic depends heavily on the location, which is ideally easily
accessible and within a given radius of a large amount of regular donors in order to meet the collection
targets. The majority of articles in this category detail a location-allocation problem [2, 4, 5, 16, 18,
21–27,29,30, 39, 42–46, 48,52] i.e. deciding the location of both mobile and fixed donation clinics, and
often also attempt to minimise the costs involved with moving clinics and transportation of blood
10
products. The next most popular area of research within this category is the planning of clinics, such
as the explicit scheduling of clinics at locations that are already assigned [2, 45]. Several articles in
this category either focus on relocating or establishing a new facility such as a blood centre or stock
holding unit [10, 16, 29, 44], whilst the comparison of effectiveness of mobile and fixed clinics is also
considered [3]. Centralisation of a regional blood service is discussed in the literature, with various
levels of centralisation considered and analysed by Osorio et al. [39].
3.1.6 Staff Utilisation
Staff Utilisation is another category lacking in research with only five articles exploring this area.
All of these propose slightly different approaches, but nearly all deal with determining the general
staffing requirements for donation clinics in order for the blood services to reach their targets for donor
satisfaction and volume of collected blood. Both Alfonso et al. [3] and Blake & Shimla [11] consider
various configurations of clinic staff and the impact of these on donor waiting times and service level,
with the former using simulation and the latter using queueing theory. Both aim to keep costs and
queues to a minimum, and along with Testik et al. [49], intend to better inform policy. However,
Testik et al. approach the problem from the perspective of donor arrivals and the effect of this on
workforce utilisation. The goal of this is to identify patterns in donor behaviour (through data mining
methods) and determine an adaptive workforce with varying numbers of staff throughout the day to
better cope with changes in donor arrivals.
Van Brummelen [13] presents an ILP model to optimally assign varying length shifts to staff to
best cope with donor arrival patterns. Van Brummelen also considers intra-day scheduling of staff
across the stations within the clinic to minimise donor waiting times and staff hours worked. This
model is based on fixed clinic sites, whereas Alfonso et al. [2] consider staff scheduling for mobile sites
also. Alfonso et al. [2] present a two-stage model in which the latter produces a staff schedule for each
clinic. However, the model does not include intra-day scheduling. Furthermore, none of the articles
consider intra-day scheduling regarding employee breaks or closure of clinic for lunch and how this
may effect donor flow. This lack of research in scheduling of staff at donation clinics is surprising,
since a lack of efficient staffing can have a massive impact on both donor satisfaction and volume of
11
collected blood.
3.1.7 Vehicle Routing
In all publications under this category, the principle aim is to minimise the distance travelled by
vehicles associated with the collections echelon, though the motivation behind this aim varies. Several
articles discuss the importance of vehicle routing in order to maximise platelet production [36, 41,
42] due to its perishable nature. S¸ahinyazan et al. [45] and Gunpinar & Centeno [24] both utilise
vehicle routing in order to maximise the number of blood donations collected, while Lodree et al. [32]
determine the optimal routes during the response phase following a large-scale disaster.
3.2 Methods
Since this review focusses on OR methods, a detailed overview about the methods and solution
approaches is given in Table 5. The approaches are categorised into Data Mining and Machine
Learning, General Statistical Analysis, Goal Programming, Heuristics, Integer Programming (includes
mixed integer programming), Queueing Theory, Qualitative, Simulation, and Stochastic Modelling.
Note that any previous literature reviews are assigned to the Qualitative category, and that the
category Statistical Analysis includes articles that incorporate probability distributions or forecasts
of certain aspects such as blood product demand or donor behaviour.
The table reveals that (mixed) integer programming is a commonly used modelling and solution
method in the field of blood collections. All of the articles that use this method, utilise it to solve an
allocation problem: either of location, appointments, staff shifts, or routes of vehicles, with the objec-
tive function mostly being to minimise costs and to maximise donor satisfaction or blood collected.
This conveys why integer programming is the most popular approach, since the effective allocation
of clinic locations, donor appointments and vehicle routes are fundamental to the success of a blood
supply chain.
Stochastic modelling is also a popular method in optimisation of collection of blood, and mostly
used alongside an integer program [5, 18, 22, 24, 26, 38, 40, 42, 43, 51, 52]. This is due to the stochastic
nature of the blood supply chain, particulary regarding both the demand of blood products and supply
12
Table 5: Methods
Methods Articles
Data Mining and Machine Learning [31, 36, 41, 49]
Statistical Analysis [2, 19, 20, 33, 34]
Goal Programming [16, 25]
Heuristics [13, 27, 32, 33, 36, 39, 41, 42, 45, 48]
Integer Programming [2, 4, 5, 7, 13, 14, 17, 18, 22–24, 26, 29, 30, 32, 36, 38, 40, 42–45, 48, 51, 52]
Queueing Theory [11, 13, 49]
Qualitative [6, 8, 9, 34, 37]
Simulation [3, 10, 13, 19, 33–35, 40, 46]
Stochastic Modelling [5, 18, 21–26, 38, 40, 42, 43, 46, 51, 52]
from donors. A significant amount of these publications present a robust optimisation approach
[5, 18, 22, 24, 25, 43, 46, 51, 52] which mostly consider the uncertainty in parameters such as donor
arrival, costs and demand. For instance, Zahiri et al. [52] propose a strategic robust possibilistic
programming model to ensure the results are still relevant and applicable over the long planning
horizon, minimising the effect of changes in parameters over time. Many of these publications focus
on disaster relief [18, 21–23, 25, 46] and utilise stochastic modelling to create a more robust blood
supply chain, to ensure effectiveness even in the midst of an emergency.
Heuristic techniques involve algorithms that seek approximate solutions quickly, and these are a fre-
quently used mathematical method within the selected literature. These are also often used alongside
integer programming methods [32, 36, 42, 45, 48], and in this case are mostly integer-programming-
based algorithms. Heuristics are also used with statistical methods to aid a primarily simulation
approach [33], as Lowalekar et al. use a gradient search-based heuristic to identify the optimum pol-
icy parameters for their model. Van Brummelen [13] uses a heuristic algorithm to allocate optimal
appointment slots throughout the day. Only one of these articles uses heuristics as their primary
method [27] - Hsieh et al. propose a solution to a location-allocation problem regarding donation
clinics, and here they use a sorting genetic algorithm to search for the Pareto set to solve the multi-
objective problem. A common theme in the utilisation of heuristic algorithms is vehicle routing
problems [32, 41, 42, 45] due to the ability of such algorithms to analyse a vast amount of possible
routes quickly; this efficiency is beneficial to many other problems within the blood supply chain, as
such problems are often very complex with a large amount of parameters and variables.
13
Simulation is widely used to analyse and optimise the blood collections process. However, some of
the articles in this category only use simulation as support or stochastic evaluation of the mathematical
model, and it is therefore not the primary method in use. For example, simulation is used as a way
of evaluating the implementation of a proposed model [13, 33, 34, 46], evaluating a current systems
performance [3], or to generate scenarios for a mathematical programming formulation [2]. The only
articles in which simulation is used as the primary method to optimise blood collections are [10,19,35].
Blake et al. [10] use simulation to determine the impact of the addition of a stock holding unit in
a given region in Canada, Lowalekar & Ravichandran [35] use simulation to compare two potential
new collection policies against each other and indeed against the current policy, and finally, Custer
et al. [19] evaluate the cost of blood per unit using simulation.
Data mining and machine learning techniques are mainly used to support other mathematical
methods. For instance, Mobasher et al. [36] and ¨
Ozener & Ekici [41] utilise clustering algorithms
to assist with vehicle routing problems. Meanwhile, Testik et al. [49] and Lee & Cheng [31] use
data mining and clustering methods to evaluate donor behaviour such as likelihood to donate and
arrival patterns at clinics. These methods provide an innovative approach to modelling of the blood
collections process as they offer the opportunity for donation clinics to be planned in alignment with
the respective donorbase of a given region, regarding the planning of location, capacity and staff.
3.3 Model Objectives
In what follows, we will break down articles which provide a mathematical model of the blood col-
lection problem into various objectives; such objectives are clearly identifiable in the case of some
methods, such as integer programming. For other methods, we identify objectives from the authors
detail on the aims of their research, and outcomes of their model. A total of 33 articles are found to
specify an objective to be maximised or minimised, with the most popular of these detailed in Table 6
whilst Table 7 displays the articles which share a common objective. The four objectives are defined
as the following:
Minimise cost: Models that aim to lessen any costs, from clinic operation to transportation
costs
14
Maximise blood: Any models that seek to maximise the total amount of blood collected.
Minimise time: This includes the minimisation of time used - either regarding a specific aspect
(such as transportation), or that of the whole blood supply chain.
Minimise distance: Any models that aim to decrease distance in some way, ranging from
distance travelled to distance between facilities.
Table 6: Popular Objectives
Objective Articles
Minimise Cost [4, 5, 11, 13, 14, 16–18, 21–23, 25, 26, 29, 30, 38–40, 42–46, 48, 51, 52]
Maximise Blood [22, 36, 42, 45]
Minimise Time [4, 5, 21–23, 26, 30, 41, 44, 48]
Minimise Distance [16, 17, 24, 46]
Both Table 6 and Table 7 clearly show that the most popular objective is minimisation of cost.
This is unsurprising since most blood services are non-profit organisations, and donors are usually
voluntary and non-remunerated. While many articles aim to minimise the cost of the collection echelon
in general, or even the whole blood supply chain, some focus on more specific costs. An example of this
is the cost of clinic operation; Salehi et al. [46] seek to minimise the costs of establishing permanent
blood centres, while Blake & Shimla [11] aim to minimise costs associated with staffing.
The minimisation of time is the second most popular objective, which is typically due to the
perishable nature of blood. This objective is often considered alongside minimisation of costs as the
two are closely linked, especially regarding transportation and staffing. Some articles present a model
which aims to minimise the length of the blood supply chain, across all echelons i.e. reduce the time
Table 7: Common Objectives
Objectives Articles
[4] [5] [16][17][21][22][23][26][30][42][44][45][46][48]
Minimise Cost 3 3 3 3 3 3 3 3 3 3 3 3 3 3
Maximise Blood 3 3 3
Minimise Time 3 3 3 3 3 3 3 3 3
Minimise Distance 3 3 3
15
blood products spend in the system, from donation to distribution at a hospital. For example, Attari
et al. and Arvan et al. [4,5] present this goal in their respective models. However, other models focus
on a specific attribute such as transportation time [21–23,30,41,48] - distance is closely related to this,
which may be the reasoning behind fewer articles detailing the minimisation of distance as a main
objective. Although, minimising distance may also concern factors other than transportation, such
as fairness in distances between blood centres and hospitals within a given region [16]. Additionally,
S¸ahin et al. [44] aim to minimise the total demand-weighted distances both from donation clinics to
blood processing centres, and from blood processing centres to hospitals.
Three articles detail the maximisation of blood collected as an objective, though from varying
perspectives. Rabbani et al. [42] wish to maximise the amount of donations that remain viable for
platelet production, while S¸ahinyazan et al. [45] aim to maximise the amount of blood collected while
optimising vehicle routing between mobile clinics. Finally, Fazil-Khalaf et al. [22] seek to collect
as much blood as possible in disaster situations. Though each blood service requires a significant
amount of blood to support demand, perhaps this objective is less popular due to overcollection of
blood leading to wastage of blood products.
Aside from the objectives listed in Table 7, various objectives were discussed individual to the
following articles [2, 7, 34, 36, 49]. Since the functional area of [7] is appointment scheduling, in this
article, Ba¸s et al. focus their objective function on balancing the production of each blood type among
days to minimise wastage of blood. Alfonso et al. [2] aim to minimise the system overtime, whilst
maximising the donor service level. Mobasher et al. [36] focus on minimising the total working time
along with the blood supplied by other regions (and thus incurring a cost). Lowalekar & Ravi [34]
concentrate on inventory-related objectives and aim to minimise both shortages and outdates of blood
products, whilst Testik et al. [49] focus on clinic operation related objectives such as maximisation of
staff utilisation and minimising donor wait.
A wide variety of objectives are covered in the selected literature, and whilst minimising costs
is of high importance to a blood service, donor satisfaction is crucial and often overlooked. As
discussed above, the minimisation of donor waiting times and service level has been considered,
though minimally. The matching of supply and demand is vital to the success and effectiveness of a
16
blood supply chain, and there is little optimisation of this considered in the literature, as it is usually
indirectly - this implies that further research which places matching supply and demand as a main
objective would be of great benefit to blood supply chains worldwide.
3.4 Planning Decision Levels
The selected articles can be categorised by the planning decision level that they discuss; namely
strategic, tactical or operational (offline or online).
Table 8: Articles by Planning Decision Levels
Planning Decision Level Articles
Strategic [2,4,5,10,14,16–23,25,29–31,33,34,39,40,43,44,46,48,51,52]
Tactical [5, 7, 13, 22, 23, 26, 27, 32, 36, 38, 41, 42, 45, 52]
Operational
Offline [2, 3, 7, 11, 13, 24, 36, 40, 45, 49]
Online [7, 13]
As described by Hulshof et al. [28], strategic planning ‘addresses structural decision making’ and
involves the decisions which help to develop and improve an organisation, more specifically in our
case, a blood service. This is therefore typically over a long planning horizon. As shown in Table 8,
the majority of the articles deal with strategic planning decisions. This is due to the vast number
of articles that deal with locational planning and collection policy, as changes to these are typically
implemented incrementally, and over a long period of time.
Tactical planning ‘translates strategic planning decisions to guidelines that facilitate operational
planning decisions’, and often involves the coordination of operations within an organisation, as
described by Hulshof et al. [28]. These decisions essentially focus on the ‘what, where, how, when
and who’ of a given process. A total of nine of the selected articles fit into this category, and these
are mainly either appointment scheduling or vehicle routing problems - these types of problems are
often solved by providing a framework for the blood service to implement, or a decision support tool.
Operational planning is typically on a short-term basis, involving the execution of a blood ser-
vice’s processes; this planning decision level is further categorised into offline and online planning.
Operational offline decisions are those that are made in advance of a process being carried out, such
17
as assigning a resource to a donor, while operational online decisions are ‘control mechanisms that
deal with . . . reacting to unplanned events’ [28] during the process. All of the articles in the opera-
tional planning category are also in the operational offline planning subcategory, with only two also
being in the operational online planning subcategory. This illustrates the need for processes to be
well-prepared and organised in advance of starting a donation clinic, but also online decisions may
need to be made regarding sickness of staff, appointment cancellations and prioritisation of donors.
Due to the complex nature of the blood supply chain, all planning decision levels are important
and necessary. However, strategic planning allows for perhaps the most significant improvements
towards a more effective and efficient supply chain, as long-term goals are able to be realised, such as
the alignment of demand and supply.
3.5 Case Studies and Implementation
A total of 32 of the articles (70%) include a case study, with real-life data from a chosen blood service,
as seen in Table 9. Several of these articles are of research motivated by a specific blood service, with
the aim of any findings being implemented if proven effective, or to inform future policy.
Table 9: Articles by Case Study Inclusion
Case Study Included Articles
Yes [2, 3, 5, 7, 10, 11, 13, 14, 18–20, 22–27, 29, 30, 33–36, 38, 39, 43–46, 49, 51, 52]
No [4, 6, 8, 9, 16, 17, 21, 31, 32, 37, 40–42, 48]
Only 14 of the articles did not present a case study within their research - note that the 4 previous
literature reviews are included in this category, and aside from these, the remaining articles propose
findings which are either tested on hypothetical data/situations or simply presented as theoretical
arguments.
The majority of selected publications presented a solution which could possibly be implemented
in real-life scenarios, yet only three of these articles specifically stated that their proposal had either
been implemented or are likely to be implemented [14,29, 40]. These findings are similar to that of
Brailsford et al. [12] who state that ‘levels of implementation for models in healthcare OR are very
small indeed’. However, this figure may not be a true representation of the implementation of the
18
current research, due to the possibility that the timeline of many projects may have ended before
implementation could be carried out, and thus the corresponding article would have no mention of
such.
4 Conclusions
This review has focussed primarily on the analysis and optimisation of the collections echelon of the
blood supply chain, as there is a distinct lack of existing reviews providing an extensive and recent
evaluation of this particular field of research. We developed a categorisation framework to distinguish
methods, functional areas and various other aspects of existing research, and this provides a clear
classification of the literature surrounding blood collections. This enabled us to identify any areas
that require further research.
Perhaps the most notable area which calls for more in-depth research to be conducted is that
of resource planning at blood donation clinics, and more specifically workforce planning. Presently,
very few publications explore this (within our literature search) with most simply analysing the levels
of required staff. Van Brummelen [13] does consider intra-day scheduling of staff and varying shift
lengths but only for fixed clinic sites. Only one model that explicitly generates a staff schedule
for mobile donation clinics [2], but this does not consider intra-day scheduling. Not only does the
assignment of staff to clinics require further research, but also the scheduling of staff throughout a
day, especially considering staff breaks and how to mitigate the effects of this on the donor waiting
times. Intra-day scheduling is of great importance as it can help to improve donor service level, reduce
waiting times and increase productivity, through utilising the workforce in alignment with varying
donor behaviour throughout the day.
Appointment scheduling is also lacking in research within optimisation of the collection of blood,
despite appointments enabling clinics to have some control over queues and donor arrivals. In the
existing literature, the research tends to focus on one aspect of appointment scheduling such as
apheresis donations, blood types, or aligning appointments with transportation of collected units.
However, appointment scheduling provides the opportunity to not only control inventory in regards
19
to volume and blood type, but also to manage the donor flow through a clinic via analysis of frequency
of appointment slots. There is much research to be undertaken that marries all of these important
aspects together.
Both the scheduling of staff and appointments have great impact on the efficiency of a clinic and
donor satisfaction. Due to donors (in most cases) being volunteers and non-remunerated, it is vital
to maximise the experience of donors as far as possible i.e. minimisation of queues, efficient service,
convenient location and appointment time, etc. Future research within this field should place donor
satisfaction at the forefront of its aims and objectives, as the success of blood supply chains ultimately
rely on the many generous donors worldwide.
Matching supply and demand is the goal of any supply chain, but achieving this goal is of utmost
importance to the blood supply chain, with failure to do so resulting in potentially critical conse-
quences. Despite this, matching the supply and demand of blood is often neglected or only indirectly
considered. Whilst many publications account for demand being satisfied, there is a significant lack
of research regarding overcollection of blood as this leads to avoidable wastage of invaluable products.
Another area which has potential for further research to be carried out is the effect of disasters on
the blood supply chain. At present, models that consider a robust and effective blood supply chain in
crisis situations are only from Asia and concerning earthquakes. However, the rest of the world would
also greatly benefit from conducting such research regarding their blood services as both natural and
human-made disasters occur worldwide and have the potential to cause devastating effects and mass
casualties.
In conclusion, as human blood is an invaluable and scarce resource which is essential to modern
healthcare, the blood supply chain is vitally important globally. Since the success of the blood supply
chain is ultimately dependent on voluntary donors in most parts of the world, further research into
the collection of donated blood is imperative to reduce wastage and shortages of blood, and increase
the effectiveness and efficiency of blood services.
20
5 Acknowledgements
The authors sincerely thank the associate editor and the anonymous referees for their careful review
and suggestions for improvement of this paper. The authors also extend their thanks to the Welsh
Blood Service for providing support and insight into the blood supply chain.
6 Funding
This work has been undertaken as part of a research project in collaboration with the Welsh Blood
Service, funded by a KESS2 scholarship, titled ‘Mathematical Modelling to Support the Redesign of
the Welsh Blood Service Supply Chain System’. Knowledge Economy Skills Scholarships (KESS) is a
pan-Wales higher level skills initiative led by Bangor University on behalf of the HE sector in Wales.
It is part funded by the Welsh Government’s European Social Fund (ESF) convergence programme
for West Wales and the Valleys.
21
References
[1] A. Ahmadi-Javid, Z. Jalali, and K. J. Klassen. Outpatient appointment systems in healthcare:
A review of optimisation studies. European Journal of Operational Research, 258(1):3–34, 2017.
doi.org/10.1016/j.ejor.2016.06.064
[2] E. Alfonso, V. Augusto, and X. Xie. Mathematical programming models for annual and weekly
bloodmobile collection planning. IEEE Transactions on Automation Science and Engineering,
12(1):96–105, 2015.
[3] E. Alfonso, X. Xie, V. Augusto, and O. Garraud. Modeling and simulation of blood collection
systems. Health Care Management Science, 15(1):63–78, 2012.
[4] M. Arvan, R. Tavakkoli-Moghaddam, and M. Abdollahi. Designing a bi-objective, multi-product
supply chain network for blood supply. Uncertain Supply Chain Management, 3(1):57–68, 2015.
[5] M.Y.N. Attari, S.H.R. Pasandideh, A. Aghaie, and S.T.A. Niaki. A bi-objective robust opti-
mization model for a blood collection and testing problem: an accelerated stochastic benders
decomposition. Annals of Operations Research, 2018.
[6] S. Bas G¨ure, G. Carello, E. Lanzarone, and S. Yal¸cindag. Unaddressed problems and research
perspectives in scheduling blood collection from donors Production Planning and Control, 29(1);
84–90, 2017.
[7] S. Bas, G. Carello, E. Lanzarone, and S. Yal¸cindag. An appointment scheduling framework to
balance the production of blood units from donation. European Journal of Operational Research,
265(3); 1124–1143, 2018.
[8] S. Bas, G. Carello, E. Lanzarone, Z. Ocak, and S. Yal¸cindag. Management of Blood Donation
System: Literature Review and Research Perspectives. Health Care Systems Engineering for
Scientists and Practitioners, 2016. doi.org/10.1007/978-3-319-35132-2 12
[9] J. Beli¨en and H. Forc´e. Supply chain management of blood products: A literature review.
European Journal of Operational Research, 217(1):1–16, 2012.
22
[10] J. Blake, K. McTaggart, and M. Hardy. Modelling a blood distribution network in the prairies
with a generic simulation framework. INFOR, 53(4):194–210, 2015.
[11] J.T. Blake and S. Shimla. Determining staffing requirements for blood donor clinics: the canadian
blood services experience. Transfusion, 54(3 Pt 2):814–820, 2014.
[12] S. C. Brailsford, P. R. Harper, B. Patel, and M. Pitt. An analysis of the academic literature on
simulation and modelling in health care. Journal of Simulation, 3(3):130–140, 2009.
[13] Samuel. P. J. van Brummelen. Bloody Fast Blood Collection Enschede: Ipskamp Printing
doi.org/10.3990/1.9789036544283
[14] G. Bruno, A. Diglio, C. Piccolo, and L. Cannavacciuolo. Territorial reorganization of regional
blood management systems: Evidences from an italian case study. Omega (United Kingdom),
2018. doi.org/10.1016/j.omega.2018.09.006
[15] J.L. Carson et al. Red blood cell transfusion: a clinical practice guideline from the AABB. Annals
of Internal Medicine, 159(1):49–58, 2012.
[16] E. Cetin and L.S. Sarul. blood bank location model: a multiobjective approach. Eur J Pure
Appl Math, 2(1):112–124, 2009.
[17] P. Chaiwuttisak, H. Smith, Y. Wu, C. Potts, T. Sakuldamrongpanich, and S. Pathomsiri. Lo-
cation of low-cost blood collection and distribution centres in thailand. Operations Research for
Health Care, 9:7–15, 2016.
[18] S. Cheraghi and S.M. Hosseini-Motlagh. Responsive and reliable injured-oriented blood sup-
ply chain for disaster relief: a real case study. Annals of Operations Research, 2018.
doi.org/10.1007/s10479-018-3050-5
[19] B. Custer, E.S. Johnson, S.D. Sullivan, T.K. Hazlet, S.D. Ramsey, E.L. Murphy, and M.P. Busch.
Community blood supply model: Development of a new model to assess the safety, sufficiency,
and cost of the blood supply. Medical Decision Making, 25(5):571–582, 2005.
23
[20] R.E. Domen, I.D. Grewal, N.V. Hirschler, and G.A. Hoeltge. An evaluation of the need for
shared blood donor deferral registries. International Journal for Quality in Health Care, 9(1):35–
41, 1997.
[21] B. Fahimnia, A. Jabbarzadeh, A. Ghavamifar, and M. Bell. Supply chain design for efficient and
effective blood supply in disasters. International Journal of Production Economics, 183:700–709,
2017.
[22] M. Fazli-Khalaf, S. Khalilpourazari, and M. Mohammadi. Mixed robust possibilistic flexible
chance constraint optimization model for emergency blood supply chain network design. Annals
of Operations Research, pages 1–31, 2017.
[23] M.R. Ghatreh Samani, S.A. Torabi, and S.-M. Hosseini-Motlagh. Integrated blood supply chain
planning for disaster relief. International Journal of Disaster Risk Reduction, 27:168–188, 2018.
[24] S. Gunpinar and G. Centeno. An integer programming approach to the bloodmobile routing
problem. Transportation Research Part E: Logistics and Transportation Review, 86:94–115, 2016.
[25] M. Habibi-Kouchaksaraei, M.M. Paydar, and E. Asadi-Gangraj. Designing a bi-objective multi-
echelon robust blood supply chain in a disaster. Applied Mathematical Modelling, 55:583–599,
2018.
[26] B. Hamdan and A. Diabat. A two-stage multi-echelon stochastic blood supply chain problem.
Computers and Operations Research, 101:130–143, 2019.
[27] C.L. Hsieh. An evolutionary-based optimization for a multi-objective blood banking supply chain
model. volume 8481, pages 511–520. Springer Verlag, 2014.
[28] P. Hulshof, N. Kortbeek, R. Boucherie, et al. Taxonomic classification of planning decisions in
health care: a structured review of the state of the art in OR/MS. Health Systems, 1(2): 129–175,
2012.
[29] D.A. Jacobs, M.N. Silan, and B.A. Clemson. An analysis of alternative locations and service
areas of american red cross blood facilities. Interfaces, 26(3):40–50, 1996.
24
[30] S. Khalilpourazari and A. Arshadi Khamseh. Bi-objective emergency blood supply chain network
design in earthquake considering earthquake magnitude: a comprehensive study with real world
application. Annals of Operations Research, pages 1–39, 2017.
[31] W.-C. Lee and B.-W. Cheng. An intelligent system for improving performance of blood donation.
Journal of Quality, 18(2):173–185, 2011.
[32] E.J. Lodree, D. Carter, and E. Barbee. The donation collections routing problem. volume 185,
pages 159–189. Springer New York LLC, 2016.
[33] H. Lowalekar, R. Nilakantan, and N. Ravichandran. Analysis of an order-up-to-level policy for
perishables with random issuing. Journal of the Operational Research Society, 67(3):483–505,
2016.
[34] H. Lowalekar and R.R. Ravi. Revolutionizing blood bank inventory management using the toc
thinking process: An indian case study. International Journal of Production Economics, 186:89–
122, 2017.
[35] H. Lowalekar and N. Ravichandran. Model for blood collections management. Transfusion, 50(12
PT2):2778–2784, 2010.
[36] A. Mobasher, A. Ekici, and O. ¨
O. ¨
Ozener. Coordinating collection and appointment scheduling
operations at the blood donation sites. Computers and Industrial Engineering, 87:260–266, 2015.
[37] A.F. Osorio, S.C. Brailsford, and H.K. Smith. A structured review of quantitative models in
the blood supply chain: A taxonomic framework for decision-making. International Journal of
Production Research, 53(24):7191–7212, 2015.
[38] A.F. Osorio, S.C. Brailsford, and H.K. Smith. Whole blood or apheresis donations? a multi-
objective stochastic optimization approach. European Journal of Operational Research, 2017.
doi.org/10.1016/j.ejor.2017.09.005
[39] A.F. Osorio, S.C. Brailsford, H.K. Smith, and J. Blake. Designing the blood supply chain: how
much, how and where? Vox Sanguinis, 113(8):760–769, 2018.
25
[40] A.F. Osorio, S.C. Brailsford, H.K. Smith, S.P. Forero-Matiz, and B.A. Camacho-Rodr´ıguez.
Simulation-optimization model for production planning in the blood supply chain. Health Care
Management Science, 20(4):548–564, 2017.
[41] O. ¨
O. ¨
Ozener and A. Ekici. Managing platelet supply through improved routing of blood collection
vehicles. Computers and Operations Research, 98; 113–126, 2018.
[42] M. Rabbani, M. Aghabegloo, and H. Farrokhi-Asl. Solving a bi-objective mathematical pro-
gramming model for bloodmobiles location routing problem. International Journal of Industrial
Engineering Computations, 8(1):19–32, 2017.
[43] R. Ramezanian and Z. Behboodi. Blood supply chain network design under uncertainties in
supply and demand considering social aspects. Transportation Research Part E: Logistics and
Transportation Review, 104:69–82, 2017.
[44] G. Sahin, H. S¨ural, and S. Meral. Locational analysis for regionalization of Turkish Red Crescent
blood services. Computers and Operations Research, 34(3); 692–704, 2007.
[45] F. G. S¸ahinyazan, B.Y. Kara, and M. R. Taner. Selective vehicle routing for a mobile blood
donation system European Journal of Operational Research, 245(1); 22–34, 2015.
[46] F. Salehi, M. Mahootchi, and S.M.M. Husseini. Developing a robust stochastic model for de-
signing a blood supply chain network in a crisis: a possible earthquake in Tehran. Annals of
Operations Research, pages 1–25, 2017.
[47] L. Shi, J. Wang, L. Stevens, P. Ness, and H. Shan. Blood safety and availability: continuing
challenges in China’s blood banking system. Transfusion, 54(2):471–482, 2014.
[48] R. Tavakkoli-Moghaddam, S. Mortazavi, S.H. Madani-Saatchi, and M. Fallah. Designing a bi-
objective model for a blood bank location-allocation problem in emergency logistics under un-
certainty. Computers and Industrial Engineering, 2016. doi.org/10.1007/s10479-017-2588-y
26
[49] M.C. Testik, B.Y. Ozkaya, S. Aksu, and O.I. Ozcebe. Discovering blood donor arrival patterns
using data mining: A method to investigate service quality at blood centers. Journal of Medical
Systems, 36(2):579–594, 2012.
[50] J. Webster and T.J. Watson. Analyzing the past to prepare for the future: Writing a literature
review. MIS Quarterly, 26(2):13–23, 2002.
[51] B. Zahiri and M.S. Pishvaee. Blood supply chain network design considering blood group com-
patibility under uncertainty. International Journal of Production Research, 55(7):2013–2033,
2017.
[52] Zahiri, B. and Torabi, S.A. and Mousazadeh, M. and Mansouri, S.A. Blood collection man-
agement: Methodology and application Applied Mathematical Modelling, 29(23-24); 7680–7696,
2013.
27
