443405

012
WMR30610.1177/0734242X12443405Ciplak and BartonWaste Management & Research

Original Article

Waste Management & Research

A system dynamics approach for

30(6) 576­–586
© The Author(s) 2012
Reprints and permission:
healthcare waste management: a case sagepub.co.uk/journalsPermissions.nav
DOI: 10.1177/0734242X12443405

study in Istanbul Metropolitan City, Turkey wmr.sagepub.com

Nesli Ciplak and John R Barton

Abstract
Healthcare waste consists of various types of waste materials generated at hospitals, medical research centres, clinics and laboratories.
Although 75–90% of this waste is classified as ‘domestic’ in nature, 20–25% is deemed to be hazardous, which if not disposed of
appropriately, poses a risk to healthcare workers, patients, the environment and even the whole community. As long as healthcare
waste is mixed with municipal waste and not segregated prior to disposal, costs will increase substantially. In this study, healthcare
waste increases along with the potential to decrease the amounts by implementing effective segregation at healthcare facilities are
projected to 2040. Our long-term aim is to develop a system to support selection and planning of the future treatment capacity.
Istanbul in Turkey was used as the case study area. In order to identify the factors affecting healthcare waste generation in Istanbul,
observations were made and interviews conducted in Istanbul over a 3 month period. A system dynamics approach was adopted
to build a healthcare waste management model using a software package, Vensim Ple Plus. Based on reported analysis, the non-
hazardous municipal fraction co-disposed with healthcare waste is around 65%. Using the projected waste generation flows, reducing
a municipal fraction to 30% has the potential to avoid some 8000 t year−1 of healthcare waste by 2025 and almost 10 000 t year−1 by
2035. Furthermore, if segregation practices ensured healthcare waste requiring incineration was also selectively managed, 77% of
healthcare waste could be diverted to alternative treatment technologies. As the throughput capacity of the only existing healthcare
waste treatment facility in Istanbul, Kemerburgaz Incinerator, has already been exceeded, it is evident that improved management
could not only reduce overall flows and costs but also permit alternative and cheaper treatment systems (e.g. autoclaving) to be
adopted for the healthcare waste.

Keywords
Healthcare waste, municipal waste, segregation, system dynamics modelling, treatment capacity

Introduction
To improve healthcare standards communities continue to invest
in various public and private healthcare facilities. However the
waste produced as a result of activities at these facilities carries a
higher potential for infection and injury than other type of waste
(WHO, 1999). Some of the problems arising from poor manage-
ment of healthcare waste may include damage to humans by
sharp instruments, disease transmitted to humans by infectious
agents, and contamination of the environment by toxic and haz-
ardous chemicals (Blenkharn, 2006). There is a particular con-
cern about infection with human immunodeficiency virus and
hepatitis viruses B and C, for which there is a high risk of trans-
mission via healthcare waste (Franka et al., 2009; WHO, 1999).
Therefore management of healthcare waste is one of the most
pressing environmental problems which regulatory agencies face
today. Likewise it is one of the research areas which has attracted
significant attention, and a great deal of research has been con-
ducted on this topic in countries such as the UK (Tudor et al.,
2005; Woolridge et al., 2005), Croatia (Marinkovic et al., 2008),
China (Yong et al., 2009), Taiwan (Cheng et al., 2010), Korea
(Jang et al., 2006), Iran (Taghipour and Mosaferi, 2009),
Mongolia (Shinee et al., 2008), Palestine (Al-Khatib and Sato
2009), Botswana (Mbongwe et al., 2008), Egypt (Abd El-Salam,
2010), Nigeria (Coker et al., 2009), Mauritius (Mohee, 2005) and
Jordan (Bdour et al., 2007).
The long-term aim of this study was to develop a system to
support selection and planning of the future treatment capacity.
This paper details the first phase involving the use of a system
dynamics model to predict the demand on healthcare waste man-
agement system in Istanbul.
School of Civil Engineering, University of Leeds, Leeds, UK
Corresponding author:
Nesli Ciplak, School of Civil Engineering, University of Leeds, Leeds
LS2 9JT, UK
Email: cn07nc@leeds.ac.uk

Downloaded from wmr.sagepub.com at FRESNO PACIFIC UNIV on January 18, 2015

Ciplak and Barton 577
Literature review
It is important to point out that the terms ‘healthcare waste’, ‘hos-
pital waste’, ‘medical waste’, and ‘clinical waste’ are still poorly
defined and often used interchangeably around the world [health-
care waste (Prem Ananth et al., 2010), medical waste (Patwary
et al., 2011), clinical waste (Hossain et al., 2011) and hospital
waste (Abd El-Salam, 2010)]. There is no universally accepted
terminology to define these terms. Therefore in this paper, health-
care waste is considered to be sourced from healthcare and
related research, diagnosis, treatment, or prevention of disease of
humans or animals.
Relatively large quantities of waste with a broad range of
compositions and characteristics are generated at healthcare
institutions. A number of studies show that a large percentage of
healthcare waste generated in these institutions could be classi-
fied as ‘domestic’ in nature. A study conducted by Olko and
Winch (2002) in England indicated that approximately 50% of
the healthcare waste generated annually could be classified as
municipal. Such studies led to much more stringent segregation
practices to be adopted especially after Hazardous Waste
Regulations came into force (Defra, 2005a). Surveys in develop-
ing countries confirm the lack of segregation, e.g. Bendjoudi
(2009) showed that the municipal waste fraction represents 75–
90% of the total Algerian healthcare waste.
A case study conducted in Istanbul by Alagöz et al. (2008)
indicated that 65% of healthcare waste generated in Istanbul is
municipal, thus only 35% of it needed special attention if it could
be successfully segregated and diverted. Although the proportion
of healthcare waste is relatively small, poor practice in segregat-
ing municipal waste from healthcare waste stream leads to the
entire bulk of the waste becoming potentially infectious. Due to
large differences in costs associated with the disposal of health-
care waste, segregation of municipal waste from healthcare waste
is an important economic factor in healthcare facilities as
acknowledged in the study conducted in typical city hospitals in
Massachusetts by Lee (2004).
It is crucial to have an understanding about the components of
the waste and their infectiousness to decide which technology
should be adopted. Although various technologies are available
for the treatment of healthcare waste, different waste categories
have to be handled differently. Prem-Ananth et al. (2010) have
stated that it is essential to look through the composition of waste
and then select the appropriate management strategies. The UK
Health Department (UK DoH, 2011) gives examples of healthcare
waste materials (HCW) for which incineration technology is the
only available solution, such as anatomical waste, chemically
contaminated samples, medicinally contaminated infectious
waste; while some other categories could be treated by alternative
technologies. Therefore, in order to differentiate these two streams
in this study, these two terms are used [Bracketed numbers are
from the List of Waste (England) Regulations (Defra 2005b)].
•• A: Incineration-only HCW, which consists of anatomical waste
(18-01-02), healthcare chemicals (18-01-06*), pharmaceuticals
Downloaded from wmr.sagepub.com at FRESNO PACIFIC UNIV on January 18, 2015
(18-01-09 or cytotoxic and cytostatic medicines 18-01-08*) for
which incineration is necessary.
•• (B) HCW such as swabs, soiled dressings, gloves (orange
bag-18-01-03*) suitable for alternative treatment (HCW
SAT), for which incineration is not a must, therefore it can be
treated by alternative treatment plants.
Materials and methods
Over the years, Istanbul has grown to become Turkey’s largest
urban centre as well as the cultural, economic, and financial cen-
tre of the country. It has a surface area of 5400 km2 with 12.5
million residents, which is almost double the population of
London. It is located on the Bosphorus Strait and extends both on
the European (Thrace) and on the Asian (Anatolia) sides of the
Bosphorus, and is thereby the only metropolis in the world that is
situated on two continents (Figure 1).
The location of establishments, organizations as well as
research institutes in Istanbul, has attracted migrants seeking
employment and education. The metropolis accommodates 200
hospitals offering acute and chronic care to patients and almost
7000 small healthcare facilities, such as healthcare centres, phar-
macies, laboratories, dental and veterinary clinics [Based on gov-
ernment statistics (Turkish Health Ministry Statistics, 2007;
Turkish Statistical Institution, 2008) and direct investigation of
the author].
The various waste-generating activities and behaviour can be
identified and used to build a waste generation model that has the
potential to respond to factors such as policy, management, train-
ing, facility provision as well as population. However modelling
the various factors and interactions is complicated, not only
because they simultaneously involve various system elements
but also because they change dynamically over time (Chaerul
et al., 2008). In addition, it is well-established that both the avail-
ability and quality of input data is limited in waste management.
This is an important challenge in producing waste projections
used for future planning purposes. One of the crucial advantages
of the system dynamics over a deterministic approach is its capa-
bility in handling assumptions and testing the impact of these
assumptions on the results where data are scarce. In this way it
enables users to identify which sort of data is essential in the first
place. Therefore, in the present study system dynamics is consid-
ered to be an appropriate tool as it is able to capture the dynamic
behaviour of the system as well as dealing with uncertainty and
assumptions about the system structures.
Healthcare waste management in
Istanbul
There is an increasing pressure on government authorities to
develop a sustainable approach for healthcare waste manage-
ment; partly to integrate with strategies aimed at pursuing a sus-
tainable society, but also to align Turkish practice to European
Union requirements. To do this, healthcare waste materials are
578 Waste Management & Research 30(6)
Figure 1.  Map of Istanbul.
regulated by the Ministry of Environment and Forestry under the
Turkish Medical Waste Control Regulation (TMWCR), which
was released for the first time in 1993 and was upgraded in 2005
(Turkish Environment and Forestry Ministry 2005).
According to the TMWCR, healthcare institutions are under a
duty of care for internal collection and storage of their waste
materials temporarily on their site. Likewise local district munic-
ipalities are legally responsible for collection, transport and dis-
posal of these wastes. All local district municipalities are
subordinate to the Istanbul Metropolitan Municipality. In prac-
tice, collection and disposal of healthcare waste is conducted by
Istac Inc, an affiliated company of the Istanbul Metropolitan
Municipality; and the district municipalities are responsible for
supervising this service. At the top of the hierarchy the Ministry
of Environment and Forestry carries out inspection of the whole
service to make sure that healthcare wastes are managed appro-
priately as stipulated in the TMWCR.
It is stated in the TMWCR that waste materials generated at
healthcare institutions are classified under three main groups,
which are municipal, healthcare and hazardous waste.
Healthcare waste is further divided into infectious waste, patho-
logical waste and sharps; while hazardous waste includes pres-
surized containers waste, waste containing heavy materials,
pharmaceutical waste, genotoxic waste and hazardous health-
care chemicals. Although some particular fractions of health-
care waste are designated as hazardous waste in the regulation,
in practice healthcare and hazardous waste materials are col-
lected together in healthcare waste bags with the exception of
sharps which are accumulated in rigid containers. As a whole
system, the healthcare institutions employ a four-container sys-
tem; red bag for healthcare and hazardous waste, black bag for
municipal waste, blue bag for recyclables and a yellow con-
tainer for sharps.
Downloaded from wmr.sagepub.com at FRESNO PACIFIC UNIV on January 18, 2015
_
+
+ -
+ births Population deaths
+ +
-
birth rate average lifetime
Figure 2.  Casual loop diagram notation portrayed with
Vensim (adapted from Sterman (2000)).
Similar to any waste management system, appropriate health-
care waste handling practices includes segregation, collection,
storage, transportation, treatment and final disposal. In Istanbul,
the municipality collects the healthcare waste produced on both of
the sides of the metropolis and transports it to the Kemerburgaz
Incinerator, which has a capacity of 1 tonne h−1 capacity and is
located on the European side. Because of the lack of capacity of the
plant, excess healthcare waste is disposed of in a landfill site with-
out any pre treatment (Eker et al., 2010). The existence of this
improper dump conflicts with the EU Landfill Directive (European
Union, 1999) which prohibits landfilling healthcare waste without
pre-treatment and so should be eliminated during the harmoniza-
tion period for Turkey to become a member of the European Union.
System dynamics modelling
Systems dynamics models are conceptual models focused on
selected parameters and their interactions. Each arrow (Figure 2)
indicates an influence of one element on another. These parame-
ters then have to be quantified as variables and their influences
have to be formulated mathematically.
Ciplak and Barton 579

Casual loop notation

Population
The word casual refers to cause-and-effect relationship. The births deaths
word loop refers to a closed chain of cause and effect. The words
represent the parameters in the system; and the arrows represent births rate average lifetime
casual connections.
Figure 2 shows the population stock example, which is fed by Figure 3.  Stock and flow diagram (adapted from Ford (1999)).
the flow of births and drained by the flow of deaths. The diagram
includes arrows linking the elements together and signing either 3. Auxiliary/constant variables are intermediate variables used
(+) or (–) on each link. These signs have the following for miscellaneous calculations.
meanings. 4. Finally the connectors (arrows) are the information links rep-
resenting the cause and effects within the model structure.
1. A causal link from one element to the other element has posi-
tive polarity (+); if the two variables in a cause-and-effect Mathematical representation
relationship change in the same direction. For example in Integral equation: Population (t) = [births (s) − deaths (s)]
Figure 2, the positive polarity on the arrow from population ds + Population (to) (1)
to births could mean that a larger population will tend to have
a greater number of deaths. It could also mean that a decrease Differential equation: d(Population)/dt = Net change in
in population causes decrease in births. population = births (t) − deaths (t) (2)
2. A causal link from one element to another element has nega-
tive polarity (–); if two variables change in opposite direc- Notation used in the model: Population = INTEGRAL
tions. In Figure 2 the negative polarity on the arrow between (births − deaths, Population (to)) (3)
deaths and population could mean that an increase in deaths
causes a decrease in population or that a decrease in deaths The Integral () function is exactly equivalent to Equation (1) and
causes an increase in population. represents that the stock (population) accumulates its inflows
(births) and discharges its outflows (deaths), beginning with an
In addition to the signs on each link, a complete loop is given a initial value of stock (population). The mathematical mapping of
sign. All dynamics arise from the interaction of just two types of a system occurs via a system of differential equations, which is
feedback loops, positive (or self-reinforcing) and negative (or solved numerically via simulation.
self-correcting) loops. The direction of sign of a feedback loop is
determined according to the direction of arrows which link the
System dynamics modelling for
parameters within the feedback loop. Specifically:
healthcare waste management
1. Positive loops tend to reinforce or amplify whatever is hap- The dynamics of the Istanbul healthcare management model was
pening in the system. In positive feedback loops an initial determined by the casual loop diagram as shown in Figure 4.
disturbance leads to further change, suggesting the presence Parameter selection and model form has been based on the
of an unstable equilibrium (for example, the population and authors’ observations in Istanbul and a review of the literature
births feedback loop in Figure 2)
2. Negative loops counteract and oppose the change. These
+
Outpatient
loops describe processes that tend to be self-limiting, pro- + demand
Population
cesses that seek balance and equilibrium. They exhibit a goal- Outpatient waste + Hospital +
generation expansion Inpatient
seeking behaviour. After a disturbance, the system seeks to +
+
demand
_ Desired bed
return to an equilibrium situation (for instance, the population + capacity
-
and deaths feedback loop in Figure 2) Growth in
hospitals
Demand for extra
bed capacity +
- Inpatient waste
+
_ generation
Outpatient +
Dynamics of stocks and flows appointment
capacity
+ Current bed
Hospitals + capacity
+
The systems dynamics models are constructed by building vari- Waste
+
segregation Regulation
ables categorized as stocks, flows, auxiliary variables, and con- + + enforcement
Healthcare waste for _ +
nectors (shown in Figure 3). alternative treatment Healthcare waste to
+ - be incinerated
+
1. Stock variables (symbolized by a rectangle) are the state varia-
bles and they represent the major accumulations in the system. + Healthcare waste
from hospitals +
2. Flow variables (valves) are the rate of the change in stock
variables and they represent those activities that fill in or Figure 4.  Casual loop diagram of HCW systems dynamics
drain the stocks. model (developed by the author).

Downloaded from wmr.sagepub.com at FRESNO PACIFIC UNIV on January 18, 2015

580 Waste Management & Research 30(6)

Initial mature Total population Initial elderly

Initial young
population population
population
Elderly migration out
Young migration Average time to Average time rate
in rate Young migration
mature to age
out rate Mature migration
out rate

Maturation Aging
Young Mature Elderly Elderly migration
Young out
population population population
migration in Young migration Mature migration out
out
(<20) (20-60) (>60)
Mature
Elderly migration in
Births migration in Elderly deaths

young deaths Mature deaths

Mature migration
Young mortality Elderly
in rate
rate mortality rate
Births per mature Elderly migration
female per year Mature Mature in rate
females mortality rate
Female
fraction

Figure 5.  Population sub-system of HCW systems dynamics model (developed by the author).

regarding the factors affecting origin, definition, composition Initial bed per head of
Annual increase in bed
per head
and weight flows of healthcare waste. The detailed breakdown of population
Bed per head
parameters in the sub-models also reflects the availability of data Average time to Hospital expansion
of population <Total population>

for Istanbul (published and additionally gathered by the author increase hospital in progress
capacity Desired bed
via information petition). For example, three age ranges of popu- capacity

lation were selected on the basis of clear differences in incidence

Rate new bed
rate in Turkey. capacity Demand for extra
bed capacity
Healthcare waste generation from healthcare facilities should Average beds
be proportional to the population, number of beds available for Growth in
per hospital

inpatients and outpatient appointment capacity of hospitals. hospitals

Current bed capacity
Initial hospitals
Segregation is required to separate healthcare waste from munic-
ipal waste; but further segregation of incineration-only HCW Hospitals
from the general HCW stream is essential in terms of allocation
of appropriate treatment facilities in place. The performance of Figure 6.  Bed inventory sub-system of HCW systems
the waste segregation process depends on the knowledge of the dynamics model (developed by the author).
hospital’s staff and visitors at the points of generation. The col-
lected waste is treated either at incinerators (incineration-only
Incidence rate per
HCW) or alternative treatment plants (HCW SAT) depending on Incidence rate per
young patient
Incidence rate per
mature patient elderly patient
its hazardous nature and then disposed of in a final disposal site. Annual young Annual mature
Annual elderly
inpatients inpatients <Elderly
<Mature inpatients
The time period of the project was assumed to be starting from <Young population population
population (20-60)> (>60)>
2007 in order to carry out a validation test (historical behaviour (<20)>

test) to compare statistical data with the results of simulation for Annual inpatient
demand
the years 2007 and 2008 and it was extended to 2040 in order to <Current bed
Average
capacity>
include the expected service lives of proposed treatment tech- Total bed
Annual inpatient inpatient stay
bed occupancy
nologies as required for civil engineering projects (30–35 years). <Hospital capacity
expansion in Average HCW
The model consists of several sub-models as shown by Figure progress> Category-1 HCW generation per bed
generation
5 (population), Figure 6 (bed inventory), Figure 7 (patient epi- Municipality
service provision
sodes; category 1), Figure 8 (patient episodes; category 2), Figure
9 (category 3), and Figure 10 (treatment/disposal). The model is Figure 7.  Category 1 sub-system of HCW systems dynamics
definitely determined when the parameters and the initial values model (developed by the author).

Downloaded from wmr.sagepub.com at FRESNO PACIFIC UNIV on January 18, 2015

Ciplak and Barton 581

Average number of
young outpatients Annual number Annual outpatient
of appointments demand Annual current
for youngs outpatient capacity
<Young population Demand for
(<20)> extra capacity
-
Outpatient Rate of new
Average number of demand appointment Outpatient
mature outpatients capacity capacity per
Annual number of hospital
appointments for Appointment capacity
matures increase in progress
<Mature Population
(20-60)> Time to response Current outpatient
capacity

Outpatient
Average number of appointments
elderly outpatients Annual number of <Hospitals>
appointments for
elderly
<Elderly population Category-2 HCW
(>60)> generation
Average HCW generation
rate per outpatient <Municipality
service provision>

Figure 8.  Category 2 sub-system of HCW systems dynamics model (developed by the author).

HCW generation rate beds per capita, which was set by State Planning Organization
Increase rate in Initial number from small HCFs of Turkish Government (2006) and released as the Ninth
small HCFs of small HCFs
Development Plan.
The demand of inpatients could either be met by building new
Category-3 hospitals or by expanding current bed capacity at existing hospi-
Number of HCW generation
tals. Both options encounter an average delay time, which stands
small HCFs
Increase in for the time difference between when the demand occurs and the
number over year government responds to this.
Municipality service provision
to small HCFs
3. Category 1, 2 and 3 (Figures 7–9).  When the healthcare
Figure 9.  Category 3 sub-system of HCW systems dynamics waste generation pattern is considered, the generation from small
model (developed by the author). healthcare institutions is more dependent on their numbers;
whereas at general hospitals, this generation pattern is based on
for the stock variables are specified. Most of the key elements in the type of treatment; acute care or chronic care (Diaz et al.,
this model are described below; 2008; Tudor, 2007). In this project ‘category 1’ is defined as
chronic care treatment at hospitals, which represents the type of
1. Population (Figure 5).  The population was divided into three the treatment that requires patients to stay in hospital over an
age cohorts: the young population who are below 20, matures extended period of time and so the waste generation is based on
between 20 and 60 and elderly population who are above 60 per bed; ‘category 2’ is for acute care in which a disease is treated
years. Rates of migration in and out and the mortality rates of for a short period of time and ‘out-patient’ is for treatment only
each of the cohorts affect the population stocks. The ‘births’ and assumes no overnight stay, so the waste generation for the
should be proportional to the average rate of births per female per latter is based only on patient numbers. While categories 1 and 2
year as well as the population of females in the mature represent the waste generation patterns based on treatment types
population. at hospitals, category 3 stands for the healthcare waste generation
from small healthcare institutions.
2. Bed inventory (Figure 6).  This sub-model was built up in Figure 7 illustrates the HCW generation from category 1
relation to inpatient (category 1) episodes to determine the patients at hospitals. Waste generation from this type of treatment
demand for extra bed capacity depending on the gap that is led by two factors: (1) in-patient demand, and (2) available bed
occurs between a current bed capacity and a desired bed capacity. In the case, where the demand of in-patients exceeds the
capacity. The desired bed capacity is based on a number of bed capacity, the waste generation is based on per in-patient.

Downloaded from wmr.sagepub.com at FRESNO PACIFIC UNIV on January 18, 2015

582 Waste Management & Research 30(6)
<Category-2 HCW
<Category-1 HCW generation>
generation>
HCW generation
from hospitals
HCW allocated to
incineration
MSW mixing
(1)
(2)
Total HCW allocated to
incineration
(3) (2)
Proportion of
MSW
segregated <Category-3 HCW
generation>
(2)
Total HCW
allocated to AT
Figure 10.  Waste segregation sub-system of HCW systems dynamics model (developed by the author).
Otherwise, the waste generation is limited by the bed capacity
and, if this is the case, the waste production is oriented by the bed
occupancy rate.
Figure 8 represents the sub-system regarding the estimation of
healthcare waste generation from category 2 type of treatment at
general hospitals. As previously stated, category 2 type of treat-
ment is based on per patient and this should be restricted by hos-
pital appointment capacity. While the waste generation from
category 2 type of treatment was simply led by the two factors (1)
waste generation rate per out-patient appointment and (2) num-
ber of appointments; the number of appointments is determined
by whether the outpatient demand is higher than the appointment
capacity. If the demand is higher than the capacity, then available
appointment capacity turns out to be a determinative factor in
waste generation, but if not, then the demand determines the
waste generation.
The ‘average number of appointments’ for each of the cohort
determines ‘out-patient demand’. Since each hospital has certain
appointment capacity, the category 2 sub-system recalls the
parameter of number of hospitals (shown as <Hospitals>) from
the hospital bed inventory sub-system. While the government
invests in building up new hospitals in the city, this leads to an
increase in bed capacity as well as the total appointment capacity.
In bridging the gap between the appointment demand and the
capacity, a delay for the feedback loop is included.
The last pattern of waste generation (category 3) is displayed
by Figure 9. The category 3 waste generation is based directly on
Downloaded from wmr.sagepub.com at FRESNO PACIFIC UNIV on January 18, 2015
Pilot scale hospitals
implemented
regulations
Schedule of
hospitals to
implement
<Hospitals> regulations
HCW allocated to
AT
Proportion of
hospitals with
Ratio of MSW mixing
implemented
with incineration-only
regulations
stream
Proportion of HCW
suitable for AT
Ratio of MSW mixing
with HCW SAT
(2)
(1)
the waste generation rate and the number of small healthcare
facilities (HCFs), whose number is re-valued over time.
4. Incidence rates (Figures 7 and 8).  In order to estimate the
number of in-patients (annual in-patient demand), ‘incidence rate’
parameters for each cohort were linked to the ‘population sub-
system’. These incidence rates represent the proportion of the
number of hospital admissions of each cohort population to the
total cohort population on an annual basis as the number of
elderly in-patients hospital admissions, for example, would not be
the same as the number of young in-patient hospital admissions.
5. Effect of the implementation of further segregation on the gen-
eration of incineration-only HCW and the HCW SAT streams
(Figure 10).  Having these two main streams segregated (this is
called ‘further segregation’) depends on how successfully the
further segregation scheme is introduced to the hospitals in the
metropolis. This effect was applied to the model by using a
lookup function, which allows customized relationships between
a variable and its causes to be defined. Lookup functions have the
same logic as an equation of y = f(x), in which the output variable
y is changed by input variable x. For this sub-system, the x vari-
able was used as the ratio of the number of hospitals implement-
ing further segregation to a total number of hospitals, and the y as
the effect of implementing further segregation on the generation
of the HCW SAT. By doing so, the output variable y was changed
by input variable x through the lookup function.
Ciplak and Barton 583
There were essentially two points to validate the function:
first, when there was no hospital implementing this segregation,
there would not be any designated HCW SAT arising – in other
words, all HCW produced would be collected together to be sent
to incineration (x = 0 and y = 0). The second validation point is
when ‘all’ hospitals implement further segregation, which makes
x equal to 1 and so y reaches the maximum value at which gener-
ated HCW SAT in the HCW stream is separated. Employing a
lookup function basically allowed predicting the waste materials
generated over the period during which the number of hospitals
is changing over time as well as the proportion of the hospitals
which implement the further segregation.
Regardless of how successfully the waste segregation is con-
ducted at hospitals it is inevitable to have some municipal solid
waste (MSW) mixed with HCW due to the logistical issues that
limit/prevent the complete segregation (e.g. lack of bins, sources
of the waste, locating waste bins). This was shown in Figure 10
by labelling the MSW mixing stream as (2), pure HCW stream
from hospitals as (1); and HCW from small HCFs as (3).
Validation of HCW model
Historical behaviour: it is one of the most common and impor-
tant tests, which is to set the inputs to the model at their historical
values and see if the outputs match history. In order to examine
whether the model can replicate the observed behaviour, the pop-
ulation and the total HCW waste generated variables were
selected. The full model was run under historical conditions
driven by the statistical data series belonging to 2007 and 2008,
as the statistical data are known for these variables in these years.
The model results give agreement with the actual values on over-
all increasing trend.
Dimensional consistency: this was checked out to determine
whether there is any inconsistency in the units of the
parameters.
Integration error tests: an integration type Euler was used in the
model as it is an acceptable integration method in the cases where
a variable time step method is used (Sterman, 2000). Since the
shortest time constant in the model is set to 1 year and standard
practice in system dynamics suggests that the integrating time
step (DT) should not be more than a quarter of the shortest time
constant in the model, the DT was initially set at one-quarter year
and the model was run. Afterwards the DT was cut into quarters
(1/16 year) and the model was run again. This made a change of
a fifth of a digit in the resultant values. Therefore DT was set as
1/16 year for the rest of the analysis.
Extreme condition test: this can be facilitated by the software, in this
case by use of the ‘reality checks’ feature in the Vensim software.
Each reality check test consists of a test input coupled to an expected
behaviour. They take the form, ‘If test input A is temporarily replaced
with a given extreme input, then behaviour B will result’.
For example, ‘births per mature female per year’ was set as
decaying over the 5 years between 2017 and 2022, after a
Downloaded from wmr.sagepub.com at FRESNO PACIFIC UNIV on January 18, 2015
considerable time, new values of ‘Young Population’ start to be
smaller than they were before. Another example used was: if
there is no hospital implementing further segregation, then no
HCW for alternative treatment appears, in response to this the
amount of HCW-to-be-incinerated peaks over the period of time.
Thirdly, if the government increases its targets on the number of
beds per capita, then there would be enough bed capacity to
accommodate all inpatients, so category 1 waste generation turns
out to be patient-based rather than bed-based.
The reality check test of Vensim only refers to behaviour; this
feature matches the requirement of a validity test as explained by
Barlas (1996): in behaviour validity tests, emphasis should be on
pattern prediction rather than point prediction because of the
long-term orientation of the model’. In other words, the emphasis
in validity tests is placed on trends rather than on the precision of
the simulated outcomes.
Even though the validity tests are important in terms of build-
ing trust in the model, it is worth emphasizing that it is impossi-
ble to correctly predict the behaviour of any model based on
observation of the system’s past (Hannon and Ruth, 1996). This
means that the output of the model should be taken as indicative
under specified scenarios only rather than as a definitive state-
ment of real future events.
Results and discussion
The section above entitled ‘Validation of HCW model’ details
the design of the system dynamics model and provides informa-
tion regarding how the model was validated. This section pro-
vides the initial waste flow results generated under a set of
assumptions regarding the segregation efficiency. Full explora-
tion of the model capabilities awaits the second phase of the
research when feedback from the decision analysis has been
completed.
A set of results of the simulation for the population, total bed
capacity at hospitals and healthcare waste generation by 2040 is
represented in Figure 11. The population increases from 12 mil-
lion in the base year to nearly 17 million at the end of the simula-
tion. The HCW arisings increases with time mainly due to the
increase in population and the investments in bed capacity, which
leads to the increase in the number of hospitals.
The healthcare waste generation is affected by the level of
waste segregation performed at the hospitals. As long as the
municipal waste fraction in healthcare waste stream is 65% in
Istanbul, the amount of healthcare waste reaches 23 000 tonnes
by 2040, which is more than twice of the annual capacity of the
Kemerburgaz incinerator.
MSW segregation
The amount of the healthcare waste can be reduced, as far as the
level of waste segregation is improved at hospitals. To analyse
how sensitive healthcare waste generation is to any improvement
in the segregation efficiency, the model was run several times
584 Waste Management & Research 30(6)

Population (million) Table 1.  HCW projections with different MSW segregation
18 levels
16
14 Years Run 4 Run 3 Run 2 Run 1
12 (tonne) (tonne) (tonne) (tonne)
10
8 2015 17 200 8800 7600 6700
6 2025 19 600 9990 8650 7600
4 2035 21 500 11 000 9500 8400
2
0
2007 2010 2015 2020 2025 2030 2035 2040
Years Table 1 gives a closer look at Figure 12 for particular years,
Hospital bed 2015, 2025 and 2035. If the fraction of municipal waste in the
healthcare waste stream is reduced from 65 to 30%, there is the
60 000
potential to avoid some 8000 t year−1 of healthcare waste by 2025
50 000
and almost 10 000 t year−1 by 2035. Furthermore a decrease from
40 000 65 to 5% results in a drop of more than 50% in the amounts of
30 000 HCW annually. This is a very important shortfall, if one consid-
20 000 ers that a saving of 8000 t year−1 is equivalent to the annual
10 000 capacity of the current incinerator. Further improvements in seg-
regation, such as reducing the MSW from 30 to 15% (difference
0
2007 2010 2015 2020 2025 2030 2035 2040
between run3 and run2) and from 15 to 5% (difference between
run2 and run1) also results in almost 1000 t year−1 less HCW to
Years
HCW (tonne) be produced.
25 000 On the other side, the segregated MSW stream inevitably
causes an increase in the amount of MSW stream at HCFs.
20 000
However, as long as the treatment of HCW requires more spe-
15 000 cialized techniques and processes, efficient segregation reduces
10 000
the treatment cost in the long term (Sawalem et al., 2009) and
provides a powerful incentive to increase the motivation of the
5 000 hospital staff, patients and the visitors in terms of disposing of
0 their waste in a right bin.
2007 2010 2015 2020 2025 2030 2035 2040
Years Further segregation
Figure 11.  Simulation of population, bed capacity and HCW Figure 13 represents the proportion of HCW SAT in the HCW
generation. stream for particular years. Although there are a few private hos-
HCW (tonne) pitals already implementing the further segregation, it is assumed
25 000 that it will take some time before all hospitals in Istanbul employ
run 4
20 000 this practice.
15 000 run 3 Figure 13 indicates that when only the healthcare waste
run 2
10 000 run 1 requiring incineration is rigorously segregated, almost 77% of
5 000 healthcare waste does not have to be incinerated, but can be
0 treated at alternative treatment plants, which can be built as mod-
2007 2010 2015 2020 2025 2030 2035 2040 ular units to treat yearly increasing arisings on a more flexible
Years
basis.
Figure 12.  Simulations with different MSW segregation
levels.
Conclusion
with different MSW segregation fractions. Figure 12 shows the The best healthcare waste management practice for healthcare
output values of the simulation which was set to run for four institutions is to prevent and minimize the generation of waste.
times; run4, run3, run2 and run1. Run4 represents current segre- However the potential for waste prevention at the point of gen-
gation practices, which is the case of 65% MSW fraction in the eration is limited because of the infectious characteristics of the
healthcare waste stream; whereas run3, run2 and run1 are the waste stream and the increased use of single-use-only disposable
runs for 30, 15 and 5% of the MSW fractions in the HCW stream, items. On the other hand, reducing healthcare waste specifically
respectively. is more likely to be achieved through appropriate waste

Downloaded from wmr.sagepub.com at FRESNO PACIFIC UNIV on January 18, 2015

Ciplak and Barton 585

HCW (tonne) HCW SAT (tonne)

tonne HCW (tonne) HCW SAT (tonne) Surveys on the generation of waste are the basis for setting
2020 18,438 14,258 77.32995904
targets for sound waste segregation and minimization as well as
25,000 2025 19,584 15,155 77.38535414
designing treatment facilities. Making the best possible estimation
2030 20,591 15,942 77.42371407 22,309
21,491 on waste generation by using any developed model is dependent
2035 21,491 20,591
16,645 77.45025666
19,584
20,000
18,438
2040 22,309 17,282 77.46835844 on first availability and then quality of input data. It is therefore
17,282
15,942
16,645 essential for healthcare institutions to record and report their treat-
14,258 15,155
15,000 ment types, patient episodes and waste generation profiles regu-
larly as well as the Turkish Health Ministry to develop a database,
10,000 which keeps this information on a standard basis. This kind of
auditing is needed to deal with uncertainties of waste management
5,000
systems by reducing the data gaps and also by improving the qual-
ity of data. Such data can then be used to further validate and
develop to form a system dynamics model over time.
0
2020 2025 2030 2035 2040 Finally the parameters incorporated into the system dynamics
Years model are not comprehensive (and in such problems never will
be), but are considered sufficient at this stage for the purpose of
Figure 13.  Proportion of HCW SAT in HCW stream. the study which was given previously. However, as and when
primary data collection improves; there are no doubt benefits to
segregation. The performance of the waste segregation process be achieved by further research into the factors influencing
depends on the knowledge of the hospitals’ staff at the points of healthcare waste generation.
generation. This brings forward the importance of training activi-
ties taking place at hospitals. Although a framing strategy was Acknowledgements
not directly addressed in this research, the system dynamics This research is related to the PhD project, which is undertaken in the
model has clearly demonstrated the deficiencies in current treat- School of Civil Engineering, the University of Leeds and sponsored
ment infrastructure. There is a potential to reduce the cost of the by the Republic of Turkey Ministry of National Education. The
treatment of healthcare waste by adopting efficient segregation authors would like to gratefully acknowledge the assistance of Dr
practices. Terry Tudor from Northampton University and Mr Andrew Hill from
Even though incineration is suitable for most types of health- Ventana Systems Inc. for providing comments on this project.
care waste and has several advantages (especially volume
reduction, fail-safe and total solution for all types of HCW), it Funding
is a costly method and might cause the release of hazardous gas This research was financially supported by the Republic of Turkey
emissions. Further healthcare waste segregation provides sub- Ministry of National Education.
stantial reductions in what would otherwise eventually end up
in the incinerator. The development of alternative treatment References
technologies for healthcare waste should be encouraged and Abd El-Salam MM (2010) Hospital waste management in El-Beheira
promoted to replace unnecessary incineration by potentially Governorate, Egypt. Journal of Environmental Management 91(3):
618–629.
more environmentally friendly treatment methods.
Al-Khatib IA and Sato C (2009) Solid healthcare waste management sta-
The predictions of this model will be used in multi-criteria deci- tus at healthcare centers in the West Bank – Palestinian Territory. Waste
sion making analysis (MCDA) to assess different potential sce- Management 29(8): 2398–2403.
narios to manage the healthcare waste of Istanbul on a wide range Alagöz AZ and Kocasoy G (2008) Determination of the best appropri-
ate management methods for the health-care wastes in Istanbul. Waste
of criteria, such as ‘treatment cost’, ‘environment performance’ Management 28(7): 1227–1235.
and ‘public health’ in the scope of a PhD project. These scenarios Barlas, Y. (1996). Formal aspects of model validity and validation in system
and criteria by which future treatment systems may be evaluated dynamics. System Dynamic Review 12: 183–210.
Bdour A, Altrabsheh B, Hadadin N, et al. (2007) Assessment of medical
by decision makers is the focus of the ongoing research. The multi wastes management practice: A case study of the northern part of Jordan.
criteria approach is a practical and feasible method for integrated Waste Management 27(6): 746–759.
assessments in terms of ranking various scenarios from different Bendjoudi Z, Taleb F, Abdelmalek F and Addou A (2009) Healthcare waste
management in Algeria and Mostaganem department. Waste Management
viewpoints of stakeholders such as metropolis administrators,
29(4): 1383–1387.
members of the government, healthcare institutions, researchers, Blenkharn JI (2006) Medical wastes management in the south of Brazil.
consultants and environmental experts. In general, MCDA, along Waste Management 26(3): 315–317.
with the results and remarks of this study, can be used as a basis of Chaerul M, Tanaka M and Shekdar AV (2008) A system dynamics approach
for hospital waste management. Waste Management 28(2): 442–449.
future planning and anticipation of the needs for investment in the Cheng YW, Li KC and Sung FC (2010) Medical waste generation in selected
area of healthcare waste management in Istanbul. clinical facilities in Taiwan. Waste Management 30(8–9): 1690–1695.

Downloaded from wmr.sagepub.com at FRESNO PACIFIC UNIV on January 18, 2015

586 Waste Management & Research 30(6)
Coker A, Sangodoyin A, Sridhar M, et al. (2009) Medical waste management
in Ibadan, Nigeria: Obstacles and prospects. Waste Management 29(2):
804–811.
Defra (Department for Environment, Food and Rural Affairs) (2005a) Hazardous
Waste Regulations. London, UK: Her Majesty’s Stationery Office.
Defra (Department of Environment, Food and Rural Affairs) (2005b) List
of Waste (England) Regulations. London, UK: Her Majesty’s Stationery
Office.
Diaz LF, Eggerth LL, Enkhtsetseg S, et al. (2008) Characteristics of health-
care wastes. Waste Management 28(7): 1219–1226.
Eker HH, Bilgili MS, Sekman E, et al. (2010) Evaluation of the regulation
changes in medical waste management in Turkey. Waste Management &
Research 28(11): 1034–1038.
European Union (1999) Council Directive on Landfill 1999/31/EC.
Luxemburg: Official Journal of the European Union Office for Official
Publications of the European Communities.
Ford A (1999) Modelling the Environment: An introduction to System
Dynamics Modelling of Environmental Systems. Washington: Island
Press.
Franka E, El-Zoka AH, Hussein AH, et al. (2009) Hepatitis B virus and
hepatitis C virus in medical waste handlers in Tripoli, Libya. Journal of
Hospital Infection 72(3): 258–261.
Hannon B and Ruth M (1996). Dynamic Modeling. New York: Springer-Verlag.
Hossain MS, Santhanam A, Nik Norulaini NA, et al. (2011) Clinical solid
waste management practices and its impact on human health and environ-
ment – A review. Waste Management 31(4): 754–766.
Jang Y-C, Lee C, Yoon O-S, et al (2006) Medical waste management in
Korea. Journal of Environmental Management 80(2): 107–115.
Lee B-K, Ellenbecker MJ and Moure-Ersaso R (2004) Alternatives for treat-
ment and disposal cost reduction of regulated medical wastes. Waste
Management 24(2): 143–151.
Marinkovic N, Vitale K, Holcer NJ, et al (2008) Management of hazardous
medical waste in Croatia. Waste Management 28(6): 1049–1056.
Mbongwe B, Mmereki BT and Magashula A (2008) Healthcare waste man-
agement: Current practices in selected healthcare facilities, Botswana.
Waste Management 28(1): 226–233.
Mohee R (2005) Medical wastes characterisation in healthcare institutions in
Mauritius. Waste Management 25(6): 575–581.
Olko P and Winch R (2002) Introducing waste segregation. Health Estate
56(10): 29–31.
Patwary MA, O’Hare WT and Sarker MH (2011) Assessment of occupa-
tional and environmental safety associated with medical waste disposal
Downloaded from wmr.sagepub.com at FRESNO PACIFIC UNIV on January 18, 2015
in developing countries: A qualitative approach. Safety Science 49(8–9):
1200–1207.
Prem Ananth A, Prashanthini V and Visvanathan C (2010) Healthcare waste
management in Asia. Waste Management 30(1): 154–161.
Sawalem M, Selic E and Herbell JD (2009) Hospital waste management in
Libya: A case study. Waste Management 29(4): 1370–1375.
Shinee E, Gombojav E, Nishimura A, et al. (2008) Healthcare waste man-
agement in the capital city of Mongolia. Waste Management 28(2):
435–441.
State Planning Organisation of Turkish Government (2006) Ninth
Development Plan. Available at: http://www.dpt.gov.tr/ing/ Law No: 877
(accessed December 2008).
Sterman JD (2000) Business Dynamics: Systems Thinking and Modeling for
a Complex World. New York, NY, USA: McGraw Hill.
Taghipour H and Mosaferi M (2009. Characterization of medical waste
from hospitals in Tabriz, Iran. Science of The Total Environment 407(5):
1527–1535.
Tudor TL (2007) Towards the development of a standardised measure-
ment unit for healthcare waste generation. Resources, Conservation and
Recycling 50(3): 319–333.
Tudor TL, Noonan CL and Jenkin LET (2005) Healthcare waste manage-
ment: a case study from the National Health Service in Cornwall, United
Kingdom. Waste Management 25(6): 606–615.
Turkish Environment and Forestry Ministry (2005) Turkish Medical Waste
Control Regulation. Ankara. Official Journal Number: 25883; Date:
22.07.2005.
Turkish Health Ministry Statistics (2007) Available at http://www.saglik.
gov.tr (accessed December 2008).
Turkish Statistical Institution (2008).Available at http://www.tuik.gov.tr
(accessed December 2008).
UK DoH (2011) Safe Management of Healthcare Waste. Version 1.0.
London, UK: UK DoH.
WHO (World Health Organization) (1999) (Pruss A, Giroult E and
Rushbrook P (eds)) Safe Management of Wastes from Health-Care
Activities. Geneva: WHO.
Woolridge A, Morrissey A and Phillips PS (2005) The development of stra-
tegic and tactical tools, using systems analysis, for waste management
in large complex organisations: a case study in UK healthcare waste.
Resources, Conservation and Recycling 44(2): 115–137.
Yong Z, Gang X, Guanxing W, et al. (2009) Medical waste manage-
ment in China: A case study of Nanjing. Waste Management 29(4):
1376–1382.