Table of Contents
Health Care Supply Chain Dynamics: Systems Design of An
American Health Care Provider
Lina K. Al-Qatawneh ®, Khalid Hafeez °, Zain Tahboub °
* School of Computing and Management Sciences,
Sheffield Hallam University, UK.
P.O.Box: 961101, Amman 11196, Jordan.
TeleFax: +(962) 6515 5167
Email: [gatawneh@ hotmail.com]
> School of Computing and Management Sciences,
Sheffield Hallam University,
Harmer Building, Howard Street, Sheffield S1 1WB, UK.
Telephone (direct line): +(44) 114 225 3438
Fax number: +(44) 114 225 3161
Email: K.Hafeez@ shu.ac.uk
° Industrial Engineering Department, School of Engineering,
University of Jordan, Jordan.
P.O.Box: 13267, Amman 11942, Jordan.
Telephone number: +(962) 6535 5000
Email
Abstract
Health care organizations supply chains are more problematic to manage compare to its
industrial counterparts. Moreover, health care organizations have very little control over the
demand of supplies. In this paper we propose an integrated system dynamics framework for
analyzing and modeling health care logistics chain. An American health care provider is used as
an example to demonstrate the implementation of various stages of our framework. Based on the
systems analyses, causal relationships were developed and an EOQ based computer simulation
model was built and tested. The analyses of the results revealed that the existing “push” type
inventory control policies are not suitable for the type of demand that is usually experienced by
the health care systems. The authors further suggest that in the health care sector, inventory
management should be based on ranking items in terms of “value” and “criticality” rather
“frequency of use” to develop appropriate stocking policy.
Keywords: health care logistics, system dynamics, causal-loop diagram, dynamic analysis, EOQ
1 Introduction
Health care is believed to be one of the largest sectors of the service industry in the world.
Offering health care and developing health services are considered fundamental national duties,
and attract lange amount of resources and corporate governance efforts around the world. Health
care constitutes one of the largest financial involvements in the United States where according to
recent statistics health care spending as a percent of GDP is 12.9 compared to: 10.4 in
Switzerland, 10.3 in Germany, 9.3 in France, 9.3 in Canada, and 6.8 in the United Kingdom
(OECD Health Data 2001). In the United States, hospital spending continues to consume the
largest portion of the health care spending -$ 371 billion in 1997 (Iglehart, 1999). It is estimated
that 30-50 percent of hospital spending is related to materials, equipment, and purchased
services, such that approximately half of this amount derives from the direct cost of acquiring
materials and services, and the other half from the cost of managing them after acquisition
(Scheyer, 1995).
Due to population demographics, governments around the globe are under constant pressure to
enhance the quality and efficiency of care within the given resources. This paper highlights some
strategies for cost reduction while maintaining a reasonable service scenario through system
dynamics analysis. Specifically, the paper discusses the implementation details of a system
dynamics framework to evaluate inventory management strategies in a children hospital’s
logistics supply chain.
2 Health Care Logistics
Health care supply chain management is considered more problematic compare to industrial
sector. This is due to the wide product range, the criticality of and the perceived need to supply
very high levels of services for each item, and the high value of products (Beier, 1995). In many
industries, fluctuations in demand can be linked to specific factors that can be controlled to some
extent (Smith, 1999). However, health care organizations have very little control over the
demand for supplies. In our view, this is due to the fact that health care industry is unique in the
volume of diverse support services required to deliver the end product - patient care.
Our experience suggests that even in a more stable demand scenario, supply chains- are difficult
to manage and design due to the complexity arising from the human, organization and
technology interactions (Hafeez et al., 1996). For example, for managing a logistics operation,
managers have to decide on inventory levels, number of warehouses, the size of the buffers,
transportation and delivery frequencies, warehouse locations, to manage critical and acute
demand fluctuations (Seppala and Holmstrom, 1995). As well as, the decision-maker has to take
into account information and material as well as cash flows, which is proving to be the major
constraint in the system. Usually the absence of a compatible operating strategy results in the
poor performance in terms of longer lead times, higher costs, alternating periods of over-capacity
followed by under-capacity, and inventory piling up at each tier of the chain. This would affect
the performance of the health care industry by stock out situation and subsequently impacting
lower level of patient care that could be very detrimental in acute treatment situations.
3 Research Methodology
Towill et al. (1992) suggest that, wherever possible, supply chain operation should be simplified
via top-down design followed by bottom-up implementation. Towill (1996) subsequently argue
that such redesign requires modeling of existing or proposed supply chains, followed by dynamic
simulation in order to synthesize alternative strategies. In this paper, the modeling and simulation
of the dynamic behavior of hospital logistics supply chains is conducted by adopting a supply
chain modeling and re-engineering methodology as described by Hafeez et al. (1996). Figure 1
illustrates the salient features of an integrated system dynamics framework that was used to
model and simulate a steel industry supply chain. The framework consists of several steps,
which go under two overlapping phases: qualitative phase and quantitative phase. Although
various stages involved are shown as sequential activities, the method is an iterative procedure,
which is represented by the feedback loops in Figure 1.
Essentially the framework decomposes the design problem into two parts: conceptual problem
and technical problem and thereby recommends using qualitative and quantitative phase to
negotiate the respective problems. The qualitative phase is related to acquiring sufficient
intuitive and conceptual knowledge to understand the structure and operation of the supply chain,
which in tum can help us in recognizing and defining the conceptual problem. Based on the
knowledge acquired from performing system analysis on the supply chain, the main variables
that have a dominant impact on the functioning and performance of the supply chain are sought
and relative cause and effect relationships and other interactions are mapped into information-
feedback loops.
The conceptual understanding sets the scene to solve the associated technical problem. The
quantitative phase concerns the development and analysis of mathematical and simulation
models. The first step towards the quantitative model building is to transform the conceptual
causal loop diagrams into block diagrams, which is verified from the concerned people.
Subsequently, the key relationships are formulated in a mathematical form using control theory
procedures, which is to be validated using field data. The model now can be subjected to a
detailed dynamic analysis to represent the time behavior of the supply chain, and suggest
improving strategies by fine tuning the existing parameters, or redesigning its structure, or
exploring different what-if scenarios.
4 The Case Study
Purchasing in the USA health sector is a relatively mature area. Usually small or medium sized
hospitals increase their buying power by forming a group purchasing organization (GPO). The
buying power of large GPOs is impressive by creating tremendous value through their scale
advantage, giving smaller hospitals the opportunity to buy like big hospitals, and big hospitals
the chance to buy like mega-chains (Chapman et al., 1998). However, there is some cost
associated in belonging to a GPO, that include the direct cost of membership - usually expressed
as an annual fee- and a certain loss of control in product selection (Scheyer, 1995). The members
of GPO’s need to meet periodically to evaluate vendor proposals, products, and performance and
to monitor how well is the group itself performing. The three biggest GPOs in USA are: Premier,
VHA/UHC, and AmeriNet. Out of over 5000 hospitals in the USA, Premier has 1850 hospitals
as its members.
Real World Supply f]
Chain
Business Objectives
Qualitative Phase
Systems Input-Output
Analysis
Conceptual Problem
Conceptual Model
Block Diagram
Formation
Statistical Techniques
Computer Simulation
Control Theory Techniques
Techniques
Lo
Verification/Validation
Technical Problem
Quantitative Phase
Dynamic Analysis.
Tune Existing What if Business
Parameters SmueilRedesion Scenarios
Figure 1: Integrated system dynamics framework for supply chain design. Source: Hafeez et al.
(1996)
As mentioned earlier we have used a children hospital from the private health care sector in the
USA as a case study. The GPO belonging to this hospital is Premier. The hospital orders its
supplies from:
1. Primary and secondary distributors: the hospital orders most of their supplies from one
primary distributor (Allegiance), and three other secondary distributors (McKesson,
Owens & Minors, and Burgen Brunswick). These distributors have their own
warehouses. They provide supplies for several hospitals for an extra charge (about 5%) to
distribute the goods according to individual hospital’s time and cost constraints. In turn,
these distributors order their supplies from product manufacturers (see Figure 2).
2. Product manufacturers: sometimes the children hospital orders its supplies directly from
product manufacturers (about 6000 manufacturers).
DAllegiance
GMcKesson
Owen & Minors
060% m Burgen Brunswick
Product
manufacturers
Figure 2: Percentage breakdown of supplier shares for the case hospital.
4.1 System Analysis
Several meetings were conducted with the materials management director of the children
hospital to gain sufficient knowledge and understanding of the structure and operation of their
supply chain. The hospital supply chain is shown in Figure 3. It includes suppliers, distributors,
health service providers and customers who are linked together via the feed forward flow of
materials and the feedback flow of information.
Supply Chain
Suppliers Distributors Service Providers Customers
Premier
(Group purchase
Primary organisation (GPO))
Product distributors:
manufacturers (Allegiance)
Secondary
distributors:
(McKesson, The childre
ec nD
Product Owens & .
manufacturers Minors, and hospital Rabients
Burgen
Brunswick)
Product |g i _ ~~ Le
manufacturers
Material flow ———_—>
Information flow -------- >
Service flow ————>
Figure 3: The overall material, information and service flow in the case hospital supply chain.
For the purpose of further analysis, we have used the following two tools:
1. Input-output analysis (IOA) (Pamaby, 1979): IOA helps to identify major systems and
the balancing of input and output flows between them (Mason-Jones et al., 1998).
2. Process flow analysis (PFA) (Towill, 1996): PFA allows defining the flow of material
through the supply chain and across functional interfaces (Mason-Jones et al., 1998).
As an example, the IOA for the central supply and main warehouse of the hospital are illustrated
in Figure 4 and Figure 5, respectively. Individual IOA diagrams as well as PFA diagrams were
then all linked together and balanced to develop an overall picture of the cash, materials,
information and service flows through the system as described in section 4.2.1.
Shortage list from
hospital wards and
departments
Supplies from main
warehouse
Inventory budget
information.
Orders from central
supply
Supplies from
suppliers
Inventory budget
information
Transport
information
———>|
tl d| Central
supply
—>
;———————_>
Supplies to hospital
wards and departments
Orders to main
warehouse
Inventory budget
updates
Central supply stock
levels updates
through computer
software
Figure 4: Input-output analysis of the Central Supply.
Main
———— | __ warehouse
+»
Supplies to central
supply
Orders to purchase
office
Inventory budget
updates
Main warehouse
stock levels updates
through computer
software
Transport
requirements
Figure 5: Input-output analysis of the main warehouse.
4.1.1 Inventory management
In most industries, the primary purpose of inventory holdings is to buffer the customer from time
lags and offer greater service levels; since having a stock out situation means diverting the
customer elsewhere and losing market share (Disney et al., 1997). However, in the health care
industry, the customer’s health is on the line, and a stock out situation may cost a life. Inventory
planning and control attempts to balance the advantages against the disadvantages of holding
stock by providing answers to two main questions: when and how much to order (Bonney,
1994). Reisman (1981, pp.1) elaborates these issues as:
Clearly, the desired or demanded service levels must be established. Optimizing between
costs and service levels requires knowledge of certain analytical techniques in order to
develop decision rules for replenishment policies. In addition, it requires an appreciation
of simulation techniques to validate and/or pretest such decision rules. Finally, a
management infrastructure to maintain and keep all systems in control is essential.
Classification of items
The items ordered by the materials management department are classified into three types: stock
items, non-stock items and special items.
¢ Stock items (fast moving items): these items are stocked at the main warehouse and
represent 98% of all items.
¢ Non-stock items (slow moving items): these items are delivered directly to the different
hospital wards and departments through the hospital receiving dock and they are not
stocked at the main warehouse. These items represent about 2% of all items.
¢ Special items: these are one-time order items.
The classification above is based upon the following criteria:
= If an item is used by the hospital 12 times/year, this is to be stocked at the main
warehouse.
= If an item -after being considered as a stock item- is used less than 3 times/year in the
following year, it will not be stocked at the main warehouse and will be considered as a
non-stock item. Otherwise it will remain considered as a stock item.
Note: A TPR card (Traveling Purchase Requisition Card) is issued for non-stock items. This card
has all the requisition information (quantity, requisition date, supplier, ...etc.). One of the
purposes of this card is to count how many times it is requested by the different hospital wards
and departments, and therefore to see if it has to be considered as a stock item or not.
4.1.2 Cash, material, information and service flows for stock items
Figure 6 illustrates the cash, materials, information, and service flows for stock items in hospital
logistics system. The different wards and departments consume supplies by importing services to
patients. This causes a decrease in the wards’ and departments’ stocks. The central supply checks
the wards’ and departments’ stock levels every 24 hours. They simply count manually what is on
shelves, and fill a prewritten list of all items in stocks. Then they top up these stocks daily to a
predetermined level from the central-supply-storage area. The central supply works as an internal
distribution system.
The central supply uses special computer software to determine its stocks’ levels. When these
levels fall below a predetermined level, an order is filled and sent to the main warehouse, which
is located one mile away from the hospital. The main warehouse then meets the central-supply
demand and checks its stocks’ levels on the software system. When the main warehouse levels
fall below a predetermined limit, an order is filled and sent to the hospital purchase office. In
response, the purchase office sends a purchase order to suppliers (primary distributor, secondary
distributors, or product manufacturers), and an electronic copy of the purchase order to the
accounts payable office (under the finance department).
Suppliers deliver supplies to the main warehouse receiving dock and send an invoice to the
accounts payable office. When the receiving dock at the main warehouse receives supplies from
suppliers, they fill a receiving note and send it electronically to the accounts payable office. Then
they deliver these supplies to the main warehouse.
The accounts payable office compares and matches the receiving note and invoice with the copy
of the purchase order and then sends payments to suppliers. Payments are usually sent 30 days
after receiving the invoice from suppliers.
4.1.3 Cash, materials, information and service flows for non-stock items
Figure 7 illustrates cash, materials, information, and service flows for non-stock items through
the children hospital logistics system. When any ward or department needs such an item, they
send a requisition directly to the hospital purchase office. In tum, the purchase office sends a
purchase order to suppliers and a copy of that order electronically to the accounts payable office.
Suppliers then deliver the items to the hospital's receiving dock and send an invoice to the
accounts payable office.
The hospital's receiving dock delivers the item directly to the ward or department that requested
that item and sends electronically a receiving note to the accounts payable office. The accounts
payable office, in tun, matches the receiving note and invoice with the purchase order and sends
payments to suppliers after 30 days of receiving the invoice from them.
‘The children hospital
Service flow
Figure 6: Cash, material, information, and service flows for stock items in the case study.
patients
‘The children hospital
patients
Service flow
Figure 7: Cash, material, information, and service flows for non-stock items in the case study.
4.2. Main Logistics Activities
Coyle et al. (1996) discussed most of the activities associated with logistics, and listed them in
Table 1. Yet, it is not necessary to place responsibility for all of these activities within the
logistics area. Haley and Krishnan (1995) argue that a logistics system that focuses on only one
particular logistical activity is too restrictive to be very useful. For example, rational
management decisions for inventories cannot be made without explicit consideration of
interrelations among activities in an overall supply system and with transportation,
manufacturing, or other processes (Clark, 1994).
Table 1: Activities associated with logistics. Source: Coyle et al. (1996)
Logistics Activity
¢ Traffic and Transportation ¢ Production planning
¢ Warehouse and storage ¢ Purchasing
¢ Industrial packaging * Customer service levels
¢ Materials handling ¢ Plant and warehouse site location
¢ Inventory control * Retum goods handling
¢ Order fulfillment ¢ Parts and service support
¢ Demand forecasting ¢ Salvage and scrap disposal
The following briefly discusses the children hospital main logistics activities that will be taken
into consideration when building the conceptual model of their logistics system. The aim of this
step is to study how the decision rules related to different logistics activities will affect the
system structure (i.e. the cash, materials, information and service flows and their corresponding
delays), and then to study in a later step the effect of system structure on system behavior over
time.
4.2.1 Approaches used for inventory management
There are three stock levels in the case hospital, namely, ward stocks, central supply (CS) and
main warehouse (MWH). A brief description of each is given below:
4.2.1.1 Wards’ stocks:
Description of policy: levels of stocks are checked at fixed intervals and sufficient supplies are
ordered to top up stocks to a predetermined level.
Fixed order interval = every 24 hours.
Desired inventory level = three day worth of usage, based on 30 days worth of data using
standard deviation. This calculation is adjusted depending on how much space is available in the
ward's storage area.
4.2.2.2 Central supply and main warehouse stocks:
Economic order quantity (EOQ) approach is used here (Coyle et al., 1996). It involves ordering a
fixed amount of product each time reordering takes place. The exact amount of product to be
ordered depends upon the product's cost and demand characteristics and upon relevant inventory
carrying and reordering costs. The stock ordering level (number of units) depends upon the time
it takes to get the new order and upon the product demand rate during that time.
2RA
*£0Q = {4
e VW
Where,
EOQ =economic order quantity (units)
R =annual rate of demand for period (units)
A =cost of placing an order ($ per order)
V =value or cost of one unit of inventory ($ per unit)
W =carrying cost per dollar value of inventory per year (% of product value)
Where reorder point: given a known lead time, multiplying lead time by daily demand
determines the reorder point.
4.2.2 Purchasing, warehousing and transportation
Purchasing, warehousing and transportation are three separate -yet related- logistics activities.
The purchasing activity is the interface between suppliers and their customers. We already
illustrated the interaction of the purchase office with other parts of the logistics system (see
Figure 6 and Figure 7). Information flow is the only type of flow that comes in and out of the
purchase office from and to the other parts. As shown in the figures, purchasing is grouped along
with other materiel-oriented functions within a single materials management department. The
purpose of this strategy is that by combining material procurement with control, many
communications lines (i.e. information flows) are shortened and purchasing policies are likely to
achieve greater strategic effectiveness. How effective this strategy is simulated as discussed in
the subsequent analysis.
Transportation is concemed with the physical movement and flow of materials between different
echelons in the supply chain. Several logistics activities have trade-off relationships with
transportation, one of which is warehousing. The decisions related to warehousing including how
many warehouses, where to locate the warehouse, what size the warehouse and so on are
affected and affects transportation-related decisions such as selecting the mode or modes of
transportation used in moving materials or developing private transportation (Coyle et al., 1996).
In this case study, as explained earlier, there are three different storage areas for stock items:
wards’ stocks, central supply, and main warehouse. Wards’ stocks are used for stocking items
used frequently when conducting services to patients. The central supply -located at the hospital
site- works as an internal distribution system to replenish the deficiencies in the wards’ stocks.
Whereas, the main warehouse -located one mile away from the hospital- is used to replenish the
deficiencies in the central supply stocks. The transport used at the hospital and at the main
warehouse- either belongs to the distributor/product manufacturer or to a third party. However,
the transportation within the hospital boundary is owned by the hospital itself.
4.3 Conceptual Model
Conceptualization is an important step in the methodology, since the mental model of the system
developed during the system analysis stage is made explicit by creating special diagrams
(Wolstenholme, 1990). Lane and Oliva (1998) summarise this step as:
The problem behavior is identified and described in a reference mode. The factors that
appear to be responsible for causing the symptoms are then identified and the
relationships between them described. These relationships are framed into information-
feedback loops that can be used, in the next phase, to model the system. The relationship
between those causal structures and the identified reference mode is called a ‘dynamic
hypothesis’ -a potential explanation of how structure is causing the observed behavior.
The feedback loops in the model are commonly diagramed using either stock-flow diagrams or
causal-loop diagrams (Albin, 1997). These diagrams are alternatively known as pipe diagrams
and influence diagrams respectively (Wolstenholme, 1990).
In this case study, the authors have used both diagrams as mediums of conceptualization. Causal-
loop diagrams -being simple to understand- were used as a tool to communicate with the
materials management director. The verified causal-loop diagram of the hospital logistics supply
chain for stock items (shown in Figure 8) is representative of the decision rules related to the
different logistics activities as adopted by the materials management department.
4.4 Computer Simulation
The verified causal-loop diagram is subsequently converted into stock-flow diagram as
illustrated in Figure 9. We have concentrated on modeling the stock items only as they represent
98% of the total requisition of the case hospital. The authors chose to develop the stock-flow
diagram using iThink Analyst. A major feature of the software is that it allows to create stock-
flow diagrams directly on the computer screen as icons, which could be opened to insert data to
create simulation models, without recourse to separate equation formulation (Wolstenholme,
1998) and (Richmond, 2001). Also the software allows creating mathematical relationships
between key variables automatically using the block diagrams.
consumption rate
Patients / CH wards boundary
(Order processing
delay “Transport delay
Order CS ddivey CSddivey CHward
Review time rio Py > te D> axicin =) sok
+ rate +
+ t+
Stodk 4+. Stock -
target ® discrepancy
(CH wards / CH CS boundary
(Order processing -
+ dday “Transport delay
Reorder level —————>_ Order MWH MWH cH
nie Py ativayte —FD—> aivaey PY css
+ + + — completion +
~ rate
+
Eoonomic ortler quantity:
Review time
CHCS/ CH MWH boundary
(Order processing -
+ dalay “Transport day
Reorder level ——————> Order ‘Suppliers Suppliers CH
rie Py > aivayrte. —TP—> aaiivay MWHstock
> aa + + — completion +
= rate
<
Eoonomic ortler quantity
Review time
(CH MWH/ Suppliers boundary
(Onder processing _
Raoueiow —$> daay “Transport delay
Suppliers +. Onkr Ddivery Ddivey Suppliers
ioe aD > | oer > rie = ——Py > ze D> axricion PP) sods
‘System boundaries Policy A as
Delay —bB—
CEE Children Hospital
CS: Central Supply
MWEH: Main Warehouse
Figure 8: The causal-loop diagram for stock items in the case study.
— va A:
ox ©
Consumption Rate
— E A:
On Transport From MWH To CS CS Stock
Cy
MWH Delivery Completion Rate
CS Order Rate CS Order Completion Rate
€s Order Backlog
CS Review Time
CS ReorderLevel CS Economic Order Quyantity
lan
MWH Delivery Rate
— a A:
On Transport From
Suppliers Delivery Rate Supeller ie MW MWH Stock
@ C5 -
Suppliers Delivery
Completion Rate
MWH Order Rate MWH Order Completion Rate
MWH Order Backlog
MWH Review Time
MWH Reorder Level MWH Economic Order Quantity
Figure 9: The stock-flow diagram for stock items in the case study.
In the causal-loop diagram, there are four stocks: wards stocks, Central Supply (CS) stock, Main
Warehouse (MWH) stock, and suppliers stock. The model in Figure 9 deals with two stocks: CS
stock and MWH stock. However, consumption of all wards were summed together and
represented in the model as consumption rate. Also, whatever comes from suppliers were
represented in the model as suppliers delivery rate.
Both CS and MWH stocks model represent a traditional EOQ inventory replenishment model of
the periodic review type. Inventory drops at a rate given by demand as a function of time and is
replenished in batches after a certain lead time, when its level reaches or has fallen below a
reorder point, represent the trigger level. The following principle assumptions usually are taken
into consideration in a simple EOQ model (Coyle et al., 1996):
¢ A continuous, constant, and known demand rate.
¢ A constant and known replenishment or lead time.
¢ The satisfaction of all demand.
¢ A constant price or cost that is independent of the order quantity or time.
¢ No inventory in transit.
¢ One item of inventory or no interaction between items.
¢ Infinite planning horizon.
¢ No limit on capital availability.
The data needed to run the model in Figure 9 are: wards consumption rate, CS and MWH order
delays, CS and MWH transport delays, CS and MWH inventory management policies (i.e.
economic order quantity (EOQ), reorder level (ROL), and the number of time units elapsing
between reviewing the inventory position). Part of the data was available from the hospital in our
case. The rest of the data were calculated according to the hospital decision rules. Data about
actual stocks level was not available. Therefore, the authors could not validate the model against
field data and see whether the model can accurately reproduce past statistical data as observed in
the real system. However, Wolstenholme (1990) argues that in system dynamics models validity
is seen as a more complex concept that centers on user confidence in the model. He adds that this
confidence stems from an appreciation of the structure of the model, its general behavior
characteristics and its ability to generate accepted responses to set policy changes. To gain this
confidence, the authors conducted several tests suggested by Wolstenholme (1990), which he
had adapted from Sterman (1984).
The model in Figure 9 was run for different items. As an example, we have used a stock item
Scalpel sterile disposable supplied by Allegiance - a primary distributor, in our simulation. A
demand pattern of this item is given in Table 2.
Table 2: Summary of demand pattern for the item used in simulation
Item description Scalpel sterile disposable 11
Item number 16205
Month/Y ear Demand
6/98 80
7/98 166
8/98 307
9/98 95
10/98 72
11/98 129
12/98 79
1/99 73
2/99 52
3/99 106
4/99 280
5/99 133
Average monthly demand | 131
Desired Quantity 41.04
Base price $ 0.7189
The required data to run the simulation model for the above item was calculated using standard
procedures and given in Appendix 1.
4.5 Dynamic Analysis
Figure 10 shows the dynamic behavior of consumption and CS stock level over time. The
consumption rate varies over a month. Therefore, CS stock depletes gradually till it reaches the
reorder level. After a time (equals to CS lead time) the CS stock is replenished by an amount
equals to EOQ. The dynamic behavior of CS stock level resembles the behavior of an EOQ
model under conditions of uncertainty, where the slope of the sawtooth varies for each order-
cycle depending on the consumption rate during that cycle.
Figure 11 shows the dynamic behavior of MWH delivery and MWH stock level over time. The
MWH delivery rate is the consumption rate of the MWH stock. As shown the MWH delivery
rate is represented as a pulse function (with a magnitude = EOQ/dt) generated by the ordering
process. The function represents the order rate and delivery rate for different batches (=-EOQ) at
certain points in time. Therefore, MWH stock level decreases abruptly in an amount equals to
EQQ at each pulse of the MWH delivery rate. For this reason, the dynamic behavior of MWH
stock level represents a square wave rather a typical sawtooth pattern representative of an EOQ
model.
This may lead to a stockout situation (i.e. if MWH stock level is less than or equals (n EOQ), and
the time between (n) successive pulses of MWH delivery rate is less than the MWH lead time.
Threfore stockout situation may be experienced first at the MWH stock and subsequently at the
CS stock.
1: Consumption Rate 2: CS Stock
25m
110
aa}
55
—
-— 1d —— |
1.00 90.75 180.50 270.25 360.00
Page 1 Days
Figure 10: The dynamic behavior of consumption and CS stock level over time.
The hospital keeps two weeks worth of safety stocks both for CS and MWH as a protection
against the uncertainties of unexpected changes in demand and an interruption in the rate of
supply. For CS the supply lead-time is small and very reliable (order delay and transport delay
are within the hospital control). Therefore, the only justification for the high safety stock level is
the protection against unexpected high demand. Y et, in case there is unexpected high demand,
which causes a stockout in CS stock, at worst cases this stockout will not be more than one day -
assuming there is no stockout or possibility of stockout in MWH stock.
1: MWH Delivery Rate 2: MWH Stock
700m
120
edd dd a ee 43 ddd
0
Cafe h 1
1.00 90.75 180,50 270.25 360.00
Page 1 Days
Figure 11: The dynamic behavior of MWH delivery and MWH stock level over time.
In overall, it was concluded that the existing inventory control policies that are based on trigger
tules (i.e. a replenishment for a batch is generated as soon as a trigger level is reached), were the
cause of the following undesirable conditions:
¢ Excessive fluctuation in stocks level (i.e. the typical sawtooth pattern generated by this
type of traditional inventory control policies).
¢ The order policies employed are non-linear (Grubbstro6m & Wikner, 1996), generating a
sequence of order impulses rather than the continuous-time order flows modeled by linear
policies.
¢ Holding high stocks levels, which was also emphasized by the Materials Management
Director of the case hospital to be one of the drawbacks of an EOQ model - especially for
slow moving items.
The authors also found that the typical inventory control policies -which typically combines
elements of both the pull and push philosophies (Coyle et al., 1996)- are not suitable for the type
of demand which usually health care systems are characterized by. As mentioned earlier at the
beginning of this paper, health care organizations have very little control over demand for
supplies. Therefore, it is suggested to shift the hospital logistics strategy from a supply-side push
orientation to a demand-side pull orientation.
5 Conclusions
This paper describes the analysis and modeling of an American health care provider logistics
supply chain using an integrated system dynamics framework. Based on the system analyses,
causal relationships were developed and a computer simulation model was built. The simulation
model was tested on a moderately fluctuating stock item. The simulation analysis reveals that
some inventory control policies that are based on trigger niles were causing the following
undesirable characteristics: excessive fluctuation in stocks level, holding high stocks level, and
employing non-linear ordering policies. Moreover, it was found that the existing “push” type
inventory control policies are not suitable for the type of demand that is usually experienced by
the health care systems. The authors further suggest that items should be ranked in terms of
value and criticality rather frequency of use to develop appropriate stocking policy. This would
allow choosing suitable level of safety stock to minimize stockouts where it is cost effective to
do so, and avoid stockouts only on essential and critical items.
References
Albin, S. (1997) Building a system dynamics model part 1: Conceptualization (D-4597).
System Dynamics in Education Project, System Dynamics Group, Sloan School of
Management, Massachusetts Institute of Technology, June 30, pp.36.
Beier, FJ. (1995) The management of the supply chain for hospital pharmacies: a focus
on inventory management practices. Journal of Business Logistics, 16(2), pp.135-73.
Bonney, M.C. (1994) Trends in inventory management. Int. J. Production Economics, 35,
pp.107-14.
Chapman, T.L., Gupta, A. & Mango, P.D. (1998) Group purchasing is not a panacea for
US hospitals. McKinsey Quarterly, 1, pp.160-5.
Clark, A.J. (1994) Multi-echelon inventory theory: a retrospective. Int. J. Production
Economics, 35, pp.271-5.
Coyle, J J., Bardi, EJ. & Langley, CJ. (1996) The management of business logistics. 6"
ed. West Publishing Company.
Disney, S.M., Naim, M.M. & Towill, D.R. (1997) Dynamic simulation modelling for
lean logistics. International Journal of Physical Distribution & Logistics Management,
27(3/4), pp.174-96.
Grubbstrom, R. W. & Wikner, J. (1996) Inventory trigger control policies developed in
terms of control theory. Int. ]. Production Economics, 45, pp.397-406.
Hafeez, K., Griffiths, M., Griffiths, J. & Naim, M.M. (1996) System design of a two-
echelon steel industry supply chain. International Journal of Production Economics, 45,
pp.121-30.
Haley, G.T. & Krishnan, R. (1995) It’s time for CALM: computer-aided logistics
management. International Journal of Physical Distribution & Logistics Management,
25(4), pp.46-60.
Iglehart, J.K. (1999) The American health care system: expenditures. The New England
ournal of Medicine, 340(1), 7” January, pp.70-6.
Lane, D.C. & Oliva, R. (1998) The greater whole: towards a synthesis of system
dynamics and soft systems methodology. European Journal of Operational Research, 107,
pp.214-35.
Mason-Jones, R., Berry, D. & Naim, M.M. (1998) A systems engineering approach to
manufacturing systems analysis. Integrated Manufacturing Systems, 9(6), pp. 350-65.
OECD Health Data (2001) Table 10: Total expenditure on health, %GDP [Internet].
Available from: <http://www.oecd.org/EN/statistics/> [A ccessed 25 February, 2001]
Pamaby, J. (1979) Concept of a manufacturing system. Intemational Jourmnal_of
Production Research, 17(2), pp.123-35.
Reisman, A. (1981) Materials Management for Health Services. D.C. Heath and
Company.
Richmond, B. (2001) An Introduction to Systems Thinking. High Performance Systems,
Inc.
Scheyer, W.L. (1995) Materials Management. In: Wolper, L.F. ed. Health care
administration: principles, practices, structure and delivery. 2"? ed. Gaithersburg,
Maryland, Aspen Publishers, pp.643-79.
Seppala, U. & Holmstrém, J. (1995) Rough modeling of logistics networks. Integrated
Manufacturing Systems, 6(5), pp.13-20.
Smith, J. (1999) Let suppliers help plan purchase needs. Hospital Materials Management,
24(1), January, pp.12-3.
Sterman, J.D. (1984) Appropriate summary statistics for evaluating the historic fit of
system dynamics models. DY NAMICA, 10(11), Winter.
Towill, D.R. (1996) Industrial dynamics modeling of supply chains. International Journal
of Physical Distribution & Logistics Management, 26(2), pp.23-42.
Towill, D.R., Naim, M.M. & Wikner, J. (1992) Industrial dynamics simulation models in
the design of supply chains. International Journal of Physical Distribution & Logistics
Management, 22(5), pp.3-13.
Wolstenholme, E.F. (1990) System enquiry: a system dynamics approach. John Wiley &
Sons Ltd.
Wolstenholme, E.F. (1998) Qualitative vs quantitative modelling: the evolving balance.
Journal of the Operational Research Society, 50(4), pp.422-8.
Appendix1: The data to run the simulation model for the Scalpel sterile disposable item
(from Table 2)
¢ The total monthly demand was averaged to get the average daily demand for that month
for the purpose of using it in the model, which was entered in the model as a graphical
function (i.e. consumption rate for certain month = average daily demand in that month).
¢ The initial value of CS stock and MWH stock = EOQ + Safety stock
=41 +61 =102
where safety stock = two weeks stock
= (2) (7) (average monthly demand/30)
= (2) (7) (131/30) = 61
¢ The time it takes CS to order the item from MWH and receive it = about 6 hours which
was divided as 3 hours ordering delay and 3 hours transporting delay.
¢ The time it takes MWH to order the item from Allegiance (their supplier for that item)
and receive it = 48 hours, which was divided as 24 hours ordering delay and 24 hours
transport delay.
¢ Reorder level =[(lead time) (average monthly demand/30)]+ (safety stock)
for CS: ROL =[(.25) (131/30)] +61=62 (where 0.25 =6 hours)
for MWH: ROL =(2) (131/30) +61 =70
¢ The report shows a desired quantity for the item = 41.04. The authors assumed that this
value is the value of the EOQ since the hospital calculates resupply for CS and MWH
based on economic order quantity EOQ. The hospital did not show us how they
calculated this value. No data were given to us about ordering cost and inventory cost.
¢ Two counters were used in the model: CS Review Time and MWH Review Time. They
represent the number of time units elapsing between reviewing the stock level for CS and
MWH respectively. For both CS and MWH inspection is done every 24 hours.
Back to the Top
