DYNAMIC IMPACTS OF PERFORMANCE BASED PAYMENT
SYSTEM ON PUBLIC HOSPITALS IN TURKEY
Tugrul Meker , Yaman Barlas
Industrial Engineering Department
Bofgazigi University
34342 Bebek Istanbul Turkey
+90 212 359 73 43
meker.tugrul@ gmail.com, ybarlas@ boun.edu.tr
Abstract
The goal of pay for performance (P4P) system in healthcare is to increase the efficiency of
healthcare resources by paying physicians and hospitals for performance. Ministry of Health in Turkey
has implemented P4P since 2004. The purpose of this study is to investigate the dynamic impacts of
P4P on the behaviors of hospitals and physicians. The model includes physicians’ interactions with
patients, the revenue pressures on physicians, and the resulting impacts on health outputs and
quality. In order to increase productivity, physicians are induced to perform more medical activities.
Physician, who experiences revenue pressure, may try to increase his/her revenue by performing more
medical activities and give less importance to quality. Resulting inadequate treatments and wrong
diagnosed patients would have negative effects on health quality. On the other hand, physicians who
do not have any revenue concerns may give the quality of healthcare absolute priority, undermining
the productivity. This tendency may result in hospital crowding and high crowding pressures on
physicians. Such conflicting pressures are included in model to investigate the impacts of P4P on
health outputs in public hospitals. Results obtained concur with our dynamic hypotheses and agree
with some of the general behaviors recently observed in Turkish healthcare.
Key words: Pay for performance (P4P), performance based payment system, health quality,
system dynamics, health modeling.
1, INTRODUCTION
The main goals of health system are to protect people’s health, to treat them if they
need any medical support and to provide better life quality. According to the OECD health
statistics, average annual growth rate in total health expenditure per capita was 4.8 % in US
between 2003 and 2009. Moreover, average growth rate in total health expenditure was 4 % in
US between 2003 and 2009 [4]. Despite the amount of money spent on healthcare, the
performance of healthcare is lower than expected. Developed and developing countries still
have to confront chronic and unsolvable problems in healthcare. Rising share of health
expenditure in GDP; long waiting times for examination, inaccessibility and disparities in
healthcare and deaths due to incorrect diagnoses and medical operations draw attention to the
efficiency in healthcare. Developed and developing countries investigate new solutions for
decreasing the costs of health meanwhile improving the healthcare quality. Consequently,
they try to implement new policies and programs for solving healthcare problems.
One of the most recently applied policies in healthcare is pay for performance (P4P) or
performance based payment system (PBPS). P4P is a common method of medical payment
system, incorporating additional payments with output and/or quality improvement. P4P
system’s aim is to increase the efficiency of healthcare resources by paying salary bonus for
increased performance. Healthcare providers usually achieve incentives for improvements in
process measures or in outcome measures. Outcome measure is the result of patient care
whereas process measure is the care that is provided [5].
Selecting process measures or outcome measures is a controversial issue. There are
advantages and disadvantages for each of these options. Process measures are easy to control
and accessible to obtain adequate information. Conversely, outcomes depend not only on
physician effort, but also on other factors beyond the control of medical professional such as
socio-economic background and environmental factors. Process measures can be defined as
time spent per examination, number of medical operations performed, number of drugs used
by patient. Outcomes can be defined as the percentage of permanent recovery, complications
due to wrong medical operations, the number of inadequate treatments etc. In order to gain
success in outcome measures, structural improvements and process improvements are needed.
In general, process measures and outcome measures are combined to get better results from
monitoring the health system, providing better health care quality, and efficient utilization of
health resources [5].
The problems in developing countries are more structural in nature as opposed to
process problems. What is meant by structural problems are organizational problems, lack of
adequate supply and high demand in healthcare, laws and policies bringing about disparities
and chronic problems in healthcare. For instance, prior to 2003, Turkish Health System was
characterized by the presence of several different public agencies funding and providing
healthcare, some vertically integrated and others relying on contractual relationships [6].
These agencies served different parts of population in different hospitals and different health
centers. Therefore, accessibility problems and disparities in healthcare might partially have
resulted from the structure of health organization itself in Turkey.
According to OECD report “Turkish Health Performance Determinants” in 2006,
physician per 1000 capita is 1.6 in Turkey whereas the OECD average for physician per 1000
capita is 3.6. Taking developing and developed country examples, it can be easily seen that
insufficiency in the number of physicians is a serious problem for Turkish Health system. The
average number of graduated physician rate for 1000 capita in Turkey is 4% per year in 2006,
whereas the OECD average is 3% per year. However; increase in birth rate and aging
population make physician graduation rate inadequate to meet the health demand.
Unfortunately, unlike developed countries, physicians may examine approximately 100
patients a day in Turkey and spend approximately four to nine minutes per examination to
meet the health demand. This tendency may result in inadequate treatments, possible
readmissions to hospitals and increase in hospital visits per year. As a result, government
decided to meet the health demand by increasing the productivity of health resources. Long
waiting lines, waiting times, inaccessibility to consultation, disparities in healthcare motivated
the ministry of health (MOH) to implement new health program: Health Transition Program
(HTP). Thus, government has initiated HTP in 2003. Government’s aim was to make the
health system more effective and efficient by improving user and provider satisfaction and
long term financial sustainability [7].
One of the reforms that government implemented as a part of HTP is PBPS. Basically,
the system awards physicians who perform more medical operations compared to the average
physician performance. The aim of this program is to increase the productivity of health
centers and physicians for meeting growing health demand.
PBPS was first implemented in pilot centers in 2003. Then, the program was extended
to cover first step public hospitals throughout the country. There were two phases of this
program. One-year implementation of PBPS in 2004 provided the participation of health
employees and health centers. Moreover, the implementation provided required infrastructure
for enabling the performance measurement of health centers and employees. Some quality
measures were tried and implemented throughout the country in 2005. Corporate performance
measurement was included in this program by the ministry of health in 2007[10].
PBPS has been in practice since 2004. This system has been applied in first, second
and third step public health centers, except university hospitals. This classification was made
by the ministry of health (MOH) in 2003[11]. First step public health centers are small health
centers such as infant health centers, village clinics and family planning centers etc. Second
step public health centers have more capability for providing more complex and complete
health service. Second step public health centers are public hospitals, social insurance
institution hospitals and other state hospitals. Third step public health centers are education
and research hospitals and university hospitals. Since February 2011, PBPS has been
implemented in university hospitals.
PBPS is used for determining how much additional payments physicians take due to
their performance. The additional payments of physicians are basically dependent on the
number of examinations, diagnostic tests and medical operations they perform. These
payments are made from the hospital’s revolving budget. Depending on hospital’s financial
performance, hospitals can allocate more money to their employees. Financial performance is
highly dependent on the revenue of hospitals. Therefore; if a hospital earns more money by
performing value added medical operations, it can pay employees more reimbursements from
revolving budget. As a result, hospitals may induce physicians to perform more examination
and medical operations in order to increase their revenue.
2. PROBLEM IDENTIFICATION AND DATA ANALYSIS
PBPS has been implemented in second and third step public health centers, except
university hospitals since 2004. When resource utilization increased in first and in second step
public health centers, government decided to implement PBPS in university hospitals in 2011.
The purpose of this study is to investigate the effects of PBPS on the performance of
second step public hospitals. These effects can be separated into three parts: the effects on
treatment quality, the effects on health costs and the effects on health productivity.
In order to understand the effect of PBPS on treatment quality, it is necessary to
characterize the health quality. While defining the health quality, it is important to take health
system as a whole and to have a whole-system perspective [12].
According to the Institute of Medicine, health quality consists of the “degree to
which health services for individuals and populations increase the likelihood of desired health
outcomes and are consistent with current professional knowledge.” [12].
According to the WHO Health Report 2006, health quality has 6 dimensions. These
dimensions are important to understand the scope of the health definition [12]. Health quality
dimensions are effectiveness, efficiency, accessibility, acceptable / patient-centered, equity
and safety.
Health policy makers should keep in mind to construct measurable quality variables to
fulfill basic health dimensions above. These variables can be waiting times for medical
treatment; time spent per examinations, treatment percentage, unit cost of medical activities
and health expenditure due to the health quality outcomes. Moreover, the effect of physician’s
revenue, health system construction, health crowding and interactions within these variables
should be taken into account for achieving desired health quality.
With respect to the health quality definitions and dimensions and variables, PBPS
should be analyzed in order to investigate the effect of the system on these variables and
interaction within health sub-systems in Turkey.
PBPS implementation in Turkey considers public health centers as revenue generating
places. The aim of health ministry is to increase the productivity, quality and efficiency in
healthcare. However these goals may contradict with each other in some ways.
With the high importance of revenue concems of hospitals and health employees,
healthcare quality may decline to second priority. Unnecessary medical operations and
examinations may be performed in order to increase hospital’s revenue. Examination
crowding in hospitals may increase to a point where health resources cannot meet. And the
gap between capacity and health demand, which is also the main problem and the main
motivation of Turkish Health System, may widen. One other result may be increases in health
expenditures which would affect the continuity of PBPS implementations. While hospital
resources have been used more efficiently since PBPS, health care expenditures have also
increased due to rising prescriptions, surgery, medical operations and examinations [7].
The following graph represents the changes in physicians’ revenue after PBPS. As it
can be seen from below; by excluding the inflation effects, physicians can increase their
revenue by performing more medical activities in first two years. Fluctuations in inflation
rates reduce the growing pattern of physician revenue in real values. These values are
obtained from Turkish Health Statistic Y earbook 2011.
7000
6000 wot
5000 FY ——Physician Revenue
4000 TL/ Month
3000 ret
—— Physician Revenue
2000 TL/ Month-2005
Real Values
1000
0 T T T 1
2002 2004 2006 2008 2010 2012
Figure 1: Physicians’ Revenue per month.
Treatment, rather than examination or surgery is an important factor for health service
quality. One way to measure treatment in healthcare is the percentage of permanent treatment
of treatable patients. PBPS may induce physicians to perform more examinations and
surgeries rather than treat patients permanently. The other way for measuring the treatment
quality is time spent per examination. With the effect of revenue concerns of hospitals and
physicians; physicians may spend less time on examination, give less attention to patients’
complaints, diagnose quickly and prescribe unnecessary medicines. Time spent per
examination in Turkey, which is very important for the correct diagnosis and permanent
treatment of patients, changes between four minutes to nine minutes. Time spent per
examination in Turkey is far lower than the OECD average. In order to increase health service
quality, time spent per examination should increase. However; with the implementation of
PBPS, time spent per examination may decrease. Reduction of time spent per examination
may be the reason of incorrect or incomplete diagnosis, unnecessary tests / analysis and
inadequate treatments. Because of the revenue concerns, numbers of medical activities
performed per year have increased since PBPS. The following graph shows number of
surgeries performed per year after PBPS implementation in healthcare. It can be interpreted
that there has been continuous increase in number of surgeries performed between 2005 and
2009.
Number of Surgeries Performed per year
4000000
3000000
2000000
1000000
0
2006 2007 2008
Figure 2: Number of Surgeries Performed per year.
However, increases in medical activities do not reflect increases in health quality
indicators. As mentioned before, there may be a negative relationship between health quality
and health productivity. Low health quality and inadequate treatments may result from more
admissions to ministry of health hospitals. This can be a reason for growing trend in Figure 3.
Number of Applications to MOH Hospital:
300000000
250000000
200000000
150000000 ae
_
100000000
50000000
0 T T T T T 1
2000 2002 2004 2006 2008 2010 2012
Figure 3: Number of Applications to MOH Hospitals.
System dynamics method is selected for understanding the dynamics of public
hospitals under PBPS. The base model will represent the dynamic impacts of the currently
implemented PBPS on second step public hospitals.
3. MODEL DESCRIPTION
Dynamic simulation model includes patients, physicians, physician’s medical
activities and performance calculation related variables. System dynamics methodology is
used in constructing the model. The motivation of this modeling study is to examine dynamic
impacts of PBPS on health outputs and quality.
In general, the main variables are patient flow related variables in hospital, salary
related variables for physicians, and revenue related variables for hospitals. Revenue related
variables are a representation of the simplified version of the complex PBPS.
For investigating patient flows in hospitals: correct diagnose rate, wrong diagnose rate,
correct treatment rate, wrong treatment rate, inadequately treated patients, surgical correction
rate and patients applying for treatment to another hospital are taken into account for building
a base stock-flow diagram that represents second step public hospital reactions to PBPS.
There are four main stocks in model: treatable patients with diagnostic, treatable
patients, inadequate treatments, chronic patients and inadequate surgeries. Treatable patients
with diagnostic represents patients who apply for medical treatment to hospital and wait for
diagnose of their health problems. Treatable patients are patients who pass diagnose process
and wait for treatment. What is meant by treatment is the treatment of special patients such as
diabetes, asthma and cancer patients. Treatment of these special patients is to resolve the
patient complaints and provide acceptable live standards and the continuity of healthcare.
Other important stock variables are inadequate treatments and inadequate surgeries.
These variables are the result of wrong diagnoses and treatments flows and affected by
various effects of time spent per examination and tests by directly or indirectly.
The main variables which affect the stocks and dynamics of the model are time spent
per examinations, number of physicians (health employee resources), hospital bed capacity,
unit performance points.
Number of patients inadequately treated is the result of inadequate treatments and
affected by time spent per examination, number of patients examined per month. Treatable
patients with diagnostic represents patients whose diagnoses are not complete and need
medical examinations and tests more than regular patients, visit hospital and apply for
treatment more than average per month.
Considering the types of medical operations performed, number of doctors in hospital
is divided into three parts in SD model: surgeon physicians, specialist physicians and
diagnostic physicians. Apart from specialist physicians, surgeon physicians also perform
surgery and can get additional payments due to the number of surgery performed per month.
Diagnostic physicians are responsible for performing tests and aiding physicians to diagnose
correctly with supplying test results.
Salary calculations for specialist physicians and surgeon physicians are pretty much
same except surgery payments to surgeons. For each month, physicians and surgeons
examine patients, perform medical operations, make hospital visits and get additional
payments due to their medical activities. If a physician performs more medical operations,
then PBPS awards him/her with more additional payments. Diagnostic physicians obtain
performance points respect to the number of tests that they perform. Salary calculations are
based on performance point calculation for month and simplified version of current PBPS.
Another important variable for PBPS is the revenue of hospital. Additional payments
from revolving budget are strictly related to the hospital’s revenue. As a result, hospitals may
induce physicians and surgeons for performing more examination and medical operations for
increasing hospitals’ income. Moreover, physicians may tend to refer more patients to
hospital care and to increase patients’ length of stay in hospital to increase the revenue of
hospital. Furthermore, surgeons may refer patients to surgery care for revenue purposes, even
if patients’ condition is not severe enough for surgery care. In addition to medical operations;
tests and analysis, which are performed in hospitals, increase hospital revenue.
Causal loop diagram of the model can be seen below. This diagram is simplified
representation of main loops of model structure. Loop’s polarity, the relationships between
variables are shown in Figure 4.
++ HosRevBudPerMonth
NumSryyPerPer\onth /
iy
NuifoPerNuesPetPer
Monit -F Month /
a \ 4, ff |
—h
Cpofiding on B: jemand f Taras Hoa con ae
a xamCrowd |
f_o4 \ |
UW \
| |
\
( jprtumbatte i + |
198 (7) Month \
‘iin \/ J
SpecActExtExamDem
Wrong
Diagnose Rate
Figure 4: Causal-Loop Diagram.
In order to examine the dynamic impacts of PBPS on health systems, a second step
public hospital is modeled. The initial conditions, the number of physicians and physician
reference revenue values are the average of second step public hospital in Istanbul.
Time horizon should extend far enough back in history to show how the problem
emerged and describe its symptoms. It should extend also far enough into to the future to
capture delayed and indirect effect of potential changes [15].
The problem/purpose of this study is the potential adverse effects of PBPS on second
step public hospitals. Time horizon for base model should be long enough to understand the
effects of PBPS. As a part of HTP, PBPS has been active since 2004.
Since, PBPS is generally based on the calculation of medical activities per month and
along term perspective is adopted, time unit of the problem was selected as month. In order to
capture real system behavior and problem dynamics, time horizon was selected as 48 months.
Time step (dt) analysis is done and time step is chosen as 1/8 month.
The model has three treatment structures: specialist physician patient’s treatment
structure, surgeon physician patient’s treatment structure and re-surgical treatment of
inadequate surgeries. The reason behind the diversity of physician patient’s treatment
structure is differences in performance point calculations of specialist and surgeons. Surgeons
can gain surgery points by performing surgeries. Inadequate surgeries stock variable is also
included in model to show dynamic behaviors of surgical treatment rate and its feedback
effects on the system.
Interactions between revenue variables and quality variables are included in model.
Physicians’ revenue concern affects TSPE. With spending less time on examinations,
physicians can increase their productivity and as a result their performance revenue.
Simplified stock model is presented below in Figure 5.
WrongSrgryFract
=
SrgryCorrFract
NumSrgryCanPerfPerMonth
Surgical
InadeSurgicalTres
Correction Rate
ary h -
= he
Se [Perfomance \
a Nr ce == {Revenue
srayneace SMSTTRE Pont ia: Cacates |
—> NumPatHospPerMonth = Structure
‘
PatRefHospFract \
TotPatExamPerMonth
= es pS
spect nesanpea erMonth _SPecTotExamDemPerMonth
Neon)
|
SpecExamCrowd |
\ TestAnaFract
fia
/ |
|
y
SpecActextexamDem
\ iagFract
|
SpecCorDi SpecCorrTreatFract SpecPhysRevPerMonth |
| * |
\\ SpecExamDemCapPerMonth |
NewSpecDiagPatFract ;
SpecWrongDiagFract-—yiSpecWronpDiagRate
|
Hospital Revenue
SpecinadeTreatPatRate
-
7 SpecReTreatRate SpecPathppOtherHospRate
SpecReTreatfract SpecWrongTreatFract
Figure 5: Simplified Version of the Stock-Flow Diagram
Surgical treatment structure includes inadequate surgeries stock variable and its flows.
Surgical correction rate and surgery patients applying to other hospital rates are outflow of
this structure below. Inflow of Inadequate Surgeries is inadequate surgically treated patients.
This flow is multiplication wrong surgery fraction and number of surgery performed per
month. Wrong surgery fraction is affected by time spent per surgery.
Two different demand sources are included in model. One is external demand and the
other one is internal demand. The internal demand is generated by visits of patients who are
still in treatment structure. Intemal demand structure is affected by extemal demand. If
internal demand increases due to decreases in health quality or health employee resources,
crowding increases as a result. Since hospital has limited capacity for medical activities,
external demand can decrease owing to increases in internal demand.
The Demand Formulation Structure is expressed in Figure 6 below:
MayS pecE xamP rod —® MaxS pecE xamCap
Specialist Treatment Structure
IntemExamDem“* |
MinS pecTSPE
MaxS pecE xtE xamC ap
Number of:
SpecExamCa
Specialist Pp "
SpecActé xt xamDem
Crowding
“
SpecP ot xtE xamDem SpecTotE xamDem
Figure 6: Internal-External Demand Structure
Internal examination demand is generated by specialist treatment structure and affects
external demand by effect of crowding in time. Moreover, 15% of external demand increases
internal demand each month.
Crowding has negative effects on treatment structure. If crowding increases, m order
to meet the demand, physicians may reduce time spent per examination (TSPE). They may
spend less time per patient and focus on increasing examination productivity to close the gap.
However, decreases in TSPE have negative effects on diagnose and treatment flows. If a
physician spend less time on TSPE, wrongly diagnosed patients and inadequately treated
patients increases meanwhile reverse effects on correct treatment and diagnose flows. Thus
negative effects of crowding result as increase in intemal demand. Due to increase in
crowding, external demand may decrease.
PBPS has complex revenue formulations. Physicians perform medical activities and in
retum, they obtain performance points. Each medical activity has unique performance points.
Physicians may prefer high incentivized points in order to increase their individual
performance. Current performance point formulation is composed of individual and group
based performance point calculation. In order to gain model simplicity and not to lose
important effects and interactions, the following formulation in Figure 7 is used.
NumSrgryPerfPerMon'
NumPatHospPerMonth
TotalPaté xamPer
Me
SpecPhy:
NumTestsAnalPerfP erMonth
SurgPhysPerfPointPerMonth
Figure 7: Performance Revenue Calculation Structure
4, MODEL FORMULATIONS
4.1. Treatment and Diagnose Rates
DiagPhysPerPerMonth
One of the main structures in the model is specialist treatment structure. This structure
represents diagnose and treatment process (Figure 8, below). External demand is the input of
this structure. SpecActExtExamDem represents actual extemal examination demand to
specialists per month. 15% of external demand is input of this structure.
Possible diagnose and treatments are included in model. This treatment structure is for
special patients. These patient’s needs, diagnoses and treatments are different from the
average patients. They visit hospital more than normal patients. Treatment of these special
patients requires more effort and time. What is meant by possible treatment or diagnose is that
physicians can only treat or diagnose as much as a percentage of their examination capacity.
PosCorrDiagRate = NumTestsPerf PerMonth * PosCorrDiagFract
SpecCorrDiagRate =
MIN(PosCorrDiagRate Treatable patients with diagnostic * SpecCorrDiagFract)
PosWrongDiagRate = NumTestsPerfPerMonth * PosWrongDiagFract
SpecWrongDiagRate =
MIN(PosWrongDiagRate Treatable patients with diagnostic *
SpecWrongDiagFract)
NewSpecDiagPatFract ‘SpecConTreatFract
Treatable
NewSpecDiagPatR ate Patients. . JccortreatRaté
SpecWrongDiagFract PosSpecConTreatRate
PosSpecWrongDiagR ate
SpecWrongTreatFract
Inadequ -—
oO: Z Treatrlen SpecWrongTreatRate
mw, il PosSpecWrongTreatRate
specinadeT PA p .
SpecPatExamferM onth 1 t SpecPatA ppO therHospFract
SpecReTreatPatFract
PosSpecReTreatPatRate
Figure 8: Treatment Structure
4.2, External-Internal Demand Formulations
External demand in the model is a function of potential extemal demand and examination
capacity of hospital. Potential external demand is constant whereas actual external demand is
function of potential external demand, examination capacity and crowding effects. The
simplified version of this formulation is expressed in Figure 9.
PotExtExamDem MaxS pecE xtE xamC ap
NormS pecE xamAvail
Graph ffAvailS pecE xt xamDem
EffAvailS pecE xt&xamDem
GraphE ffS pecCrowdE xt SpecExamC rowd HIS pi xi
ExamDem
EffS pecC rowdE xtExamDem SpecActéxtéxamDem
Figure 9: Extemal-internal Demand Structure
One of the main effects in this formulation is effect of availability on actual external
examination demand. Patients can generate examination visits to hospital with respect to its
examination capacity. In order to include this assumption into the model, the following
equations are used.
NormSpecExamAvail = PotExtExamDemSpec/MaxSpecExtExamCap
Ef f AvailSpecExtExamDem = F (NormSpecExamAvail)
It is assumed that specialist examination crowding has a negative effect on external
demand. If the crowding is far higher than average, than the patients whom apply for medical
service cannot get any treatment or examination. Lack of health service induces patients to
seek other hospitals to fulfill medical needs.
SpecExamCrowd = TotSpecExamDemPerMonth/SpecExamCapPerMonth
Ef fSpecCrowdExtDem = F (SpecExamCrowd)
SpecActExtExamDem = Ef f AvailSpecExtExamDem * MaxSpecExtExamCap *
DelEf fSpecCrowdExtDem
The following graph displays the relationship between examination crowding and
external demand. As it can be seen from the graph above, there is a negative relationship
between crowding and external examination demand. If examination capacity is higher than
demand then low crowding stimulate more potential patients to apply to hospital for medical
services.
[raph Loop -GapheFpesCromEaDemand [raph Lookup - GraphefavalspecAcDem
| my |
gt pm son] gy a a:
| = 1 aa a 4 1
or joz—[o =
1 (0.95 [Orn Ot o
me fara ae [oO
re fae ° a9 [amt
rate jaro a [ot
ant frss a2 [ez
seam a3 [07
am fase aa [0a
pr fase 05 [08
sie paar Yi a fae — | ca
New S| 03 x | ue a q
I I I I
NeinJO5 —s[.og 4-087 %maw{2___>|_ Rest Seaing Xnin|03r|e01844 yet Xai] Reset Sosing
oc | _chstons| ceaalPorts| Castel) charleense| AetCu| _ cancel Ok | _OeavPite| _CewAlPote| Cue] _CenRefnene | eu] Coma
Figure 10: Effect of Crowding on External Demand
4.3. Time Spent Per Examination
Time spent per examination is one of the most important variables in the model. In
order to provide adequate and quality diagnose/ treatment, time is vital. If physicians spend
more time on examinations, they can spend more time for taking information about patient’s
complaints. With the aid of better knowledge and understanding of patient’s complaint,
physicians may make more accurate diagnoses and adequate treatments.
It is assumed in the model that time spent per examination is affected by physicians’
and hospitals’ revenue concerns and medical activities’ crowding. The formulation of this
variable in model is combination of additive-multiplicative effect formulation. If physician’s
revenue is lower than the reference, than the physician may feel a pressure and obligation to
produce more examinations to get more performance points. If physician’s revenue is higher
than the reference, than the physicians may focus on making more accurate diagnoses and
correct treatments. The effect of physician revenue has greater effect on TSPE than hospital
revenue concern.
Hospital revenue is important to describe the effects on TSPE. Hospital revenue is
strictly related to medical operations that perform in hospital. Thus, hospitals which have
lower revenue than average, feel bankrupt pressure on themselves. Their managers seek ways
to increase hospital revenue. Thus, they induce physicians to spend less time on examinations
to increase productivity and examinations.
(Gag Leck - GaphtspecRerTSPE
| vin |
Fin |) [lop Our vinaz
fol iat naz cel = —
Lee eel) [pare |oes
Dra
Tee EE Dart loves facars?
loony facasse
losses _[cocaor 5
Frost [coeesr
aa [ace fra7_ looser
frase fare frat [a0se42
faa [Oa78 frss__[aoaear
raat (076 hima [aosara
fre3e_ooasee
bd ‘min: (1.931 (0.03863 a ‘Yin:
New A 035 ¥ New 02 +
=| I rT I
XminJS —_v]O7416 401373 Xan |.8 >| Reset Scalng Kmrie|05 x feO.7599 ye O1276 XX mae|2 | Reset Scalng
OK | ClearPoits | ClerAlPoins | CursRet| CearRefererce| RetsCul _Carce OK | Cha Poin | CeaiAlfFonts | CuRet| CleerReforence | Ref>Cu| Cancel
Figure 11: Effect of Specialist and Hospital Revenue on TSPE
Time spent per examination is also affected by crowding. Crowding is a function of
examination demand and examination capacity. There is a negative relationship between
crowding and TSPE. If crowding is higher than reference, physicians feel pressure of meeting
the examination demand. Thus, they spend less time on examination; give second priority to
adequate treatments. By decreasing TSPE, physicians can examine more patients and gain
better performance revenue.
‘Output =
ors +. zn,
0.189
(01693 ie Dnt
arse
acaaie
0.03114
0.06974
465
0.1991
12486
0.2825 | ‘Y-min:
New. al 03 >
| I I
XminfOS fed 0 Xmael15 >| Reset Scaing
is Clear Points Clear All Points Cur>Ret Clear Reference | Ref->Cur Cancel
Figure 12: Effect of Specialist Examination Crowding on TSP
The resulting equation is:
SpecTSPE =
NormalSpecTSPE * (1 + DelEf f HosRevSpecTSPE + DelEf fSpecExamCrowdTSPE +
DelEf fSpecRevTSPE)
4.3.1. Effects of TSPE on Correct and Wrong Diagnose Rates
TSPE has important effect on correct and wrong diagnose rate. If physicians spend
more time per examination, they may diagnose patients more accurately and treat patients
more correctly. There is a positive relationship between TSPE and correct diagnose rate and
negative relationship between TSPE and wrong diagnose rate. The following graphs show the
relationship between TSPE and these variables.
{ Gap Lookup -GrephperTSPConBegrat Graph Lookup - GraphEfiSpecWrongDiagFract
Pin | | Frin_|
Inpui__ulput Aig Input__Output Yana
iar [cae | jr ea fore =| a be
(05633 [0.2561 5856 (0.2956 ‘
06183 0.2167 - mn . ri ia __| pm foes2i [0.285 Dim
ace [arr (ovaae [02561 \
ova fare jesra _[nvaea
em ra joes nasos
aes [aera to
1 [o Free foosas
Ca ce | oe ire jor
ia pee OF faa [ose — \
[1.262 008728 | “atin: jise4 o1g78 +) Yerire
te ye jas] imi 2
a I I
XminfoS fed vo marl | Reset Sina Xmin|OS w]e. ye 0211 maxi _e| AecetSoaina
OK | CesPenis| GeaatPotts| Cua | Che Relerce | AesCu| Con Ok | CeaPonis| CeaatPeris| CurRet| _Cearetrone | RefsCu] Cancel
Figure 13: Effect of TSPE on Correct and Wrong Diagnose Rate
4.4. Hospital Revenue Formulation
In current performance revenue formulation in Turkey, hospital revolving budget has a
complex calculation method. Hospitals can distribute only 40% of their income to physicians
when they achieve the best performance points according to PBPS. In order to model only
related aspect of real system, the revolving budget formulation is simplified. Reference
revenue is added into model. And this variable is calculated by hospital resources, private
sector second step public hospital base revenue and public second step hospital base revenue.
The following diagram demonstrates the relationship between these variables.
number of number of
numberof —_ diagnostic surgeons
jh
physicians —_ physicians evbernedd
evPerS pec
nl Ne RevPerDiagPhys pi Nueawor
poe
NormHospRevBudFrac
big ng
Pri
Sh Bans
a
we
HospRevBudPerMonth
HospRevPerMonth
Figure 14: Hospital Revenue Formulation
BaseSecStepPubHospRev = Number of physicians * RevPhysSpec +
Number of surgeons * RevPerSurg + Number of diagnostic physcians *
RevPerDiagPhys + Number of Beds * RevPerBeds
GoalSecStepHospRev =
BaseSecStepPubHospRev * 0.2 + PrivSecStepHospRev * 0.2 + HospRevPerMonth *
0.6
PrivSecStepHospRev = BaseSecStepPubHospRev * 1.5
NormHospRevPerMonth = HospRevPerMonth/RefHospRev
HospRevBudPerMonth = HospRevBudFract * HospRevPerMonth
Goal reference formulation for hospital revenue is the weighted average of hospital
current revenue, base public hospital revenue and private hospital revenue. Weight of hospital
revenue is higher than other revenue variables, because it represents average of all second step
public hospitals and it has greater effect of calculation of goal hospital revenue.
5. MODEL VALIDATION
The aim of model validation is to assure that the model is an acceptable description of
the real system behavior with respect to the dynamic problem [16]. Model validation is
executed in two steps: structure and output behavior testing.
5.1. Structure Validity
Structure test is to check whether the structure of a model is a meaningful description
of the real relations that exists in the problem [16]. There are two types of structure tests:
direct structure tests and indirect structure tests. In the model all parameters and variables
have real-life counterparts. The model is dimensionally consistent in all equations.
One typical indirect structure testing is extreme condition testing. In order to check
whether the model is valid or not, some extreme conditions are simulated. One of model
inputs is extemal demand, and external demand is affected by potential demand. If potential
external demand is zero, then there is no input to the treatment structure. Thus, total demand
(including internal one) decreases due to the lack of external demand as expected.
Another extreme-condition test is applied on the effect of physician revenue on TSPE.
When examination capacity is higher than demand, there is no decrease in TSPE due to
revenue concer, as expected. In addition to this, an extreme condition test is applied to health
resources. When there is only one physician, all treatment stock levels decrease drastically as
expected. These and other extreme condition tests are consistent with real life information.
5.2. Base Run
As seen in Figure 15, inadequate treatment stock reaches its new high equilibrium
level in 30 months. Inadequate treatments increase due to decreasing health quality indicators
like TSPE. Treatable patients and treatable patients are stable due to slow changes in flow
variables of Figure 16.
As it can be seen in Figure 17, time spent per examination and surgery decreases
within 30 months. This is a result of revenue concerns. Physicians tend to spend less time on
medical activities to increase their revenue.
In Figure 18, it can be seen that PBPS has negative impact on quality indicators. Due
to spending less time on medical activities, correct treatment and diagnose ratios decrease as
expected.
SpecTreatable patients
8,000
6,000
= 4,000
2,000 L533
0
0 6 12 18 24 30 36 42 48
Time (Month)
Treatments : BaseRun
Spec?’ : BaseRun
SpecT reatable Patients with Diagnostic : BaseRun
Figure 15: Treatment Structure-Specialist Physician Main Stocks
Specialist Flow
0 6 12 18 24 30 36 42 48
Time (Month)
SpeclnadeTreatPatRate : BaseRun
SpecComDiagRate : BaseRun
SpecW BaseRun
SpecComTreatRate : BaseRun
SpecReTreatment Rate : BaseRun
Figure 16: Treatment Structure 2-Specialist Physician Main Flows
Time Spent Per Examinations and Tests
ood
0 6 12 18
24 30 36 42 48
Time (Month)
SpecTSPE : BaseRun.
SurgT SPE : BaseRun
TimeSpentPerTests : BaseRun
Figure 17: Time Spent Per Examination and Tests
Quality Output-1
08 Pag
in
pest epee TT
0 6 12 18 24 30 3642 48
Time (Month)
SurgConDiagRatio : BaseRun
SpecComDiagRatio : BaseRun
SurgConTreatRatio : BaseRun
SpecConTreatRatio : BaseRun
Figure 18: Correct-W rong Diagnose and Treatment Ratios
5.3. Behavior Validity
Behavior pattern tests are designed to measure how accurately the model can
reproduce the major behavior pattems of the real system [16]. Real data is limited for our
study. There is no available data for TSPE or other quality indicators. But we can guess the
real system behavior by looking into the patterns in other health over the years since PBPS
implementation.
According to the model assumptions, there is a negative relationship between
physician’s revenue concerns and health quality. Since the physician’s revenue is lower than
the reference revenue, physicians spend less time per examination and perform more
examinations and medical operations to improve their revenue. As a result, physicians’
revenue is expected to increase after P4P. In Figure 19, it can be seen that, the model and the
real data is well-matched in first year. However; owing to the continuous changes in
government policies and operations’ pricing, physicians’ revenue decreases in second year.
Between 2007 and 2009, the real-life behavior is stable just like model’s behavior.
With increases in health resources’ capacity, medical operations performed per year
increase after P4P. In Figure 20, Figure 21 and Figure 22, the model’s behavior and the real
data are well-matched. The real data for medical operations are taken from a second-step
public hospital in Istanbul.
5000
4000 | ea —
3000
—¢—Real Data
2000 =" Model
1000
0 1 1 1 r 1
2004 2005 2006 2007 2008 2009 2010
Figure 19: Physicians’ Revenue per month (MOH, Statistical Y earbook 2011)
300000 +
250000 -
150000. —$ $$ —l- Model
t= Real Data
100000 -
50000 -
04 T
ie} 20 40 60
Figure 20: Number of patients examined per month
120000 -
100000 ~
80000 pes
60000 ——Real Data
=i Model
40000 +
20000 -
04
i) 20 40 60
Figure 21: Number of tests performed per month
7000 ~
6000 - -—
5000 -
4000. -—<— S$
Model
3000 +
i Real Data
2000 -
1000 -
o 4
2004 2006 2008 2010
Figure 22: Number of surgeries performed per month
6. SCENARIO ANALYSIS
6.1. High Incentive for Performance, Adequate Demand, Adequate Physicians
In this scenario, government’s primary goal is to improve health service quality.
Adequate health budget gives MOH flexibility to carry out their performance program.
In order to reach this goal, government first increases health employee resources. Main
expectation is to meet the health demand and increase health productivity. But since there is
abundant demand, increases in health employee’s numbers would not close the gap between
health demand and capacity. Moreover, increases in physicians’ revenue also increase their
goal revenue in time. Thus, crowding and revenue concerns push physicians to decrease TSPE
and to give second priority on healthcare quality. The policy does not yield the desired
outcomes, due to compensating feedback loops in the system.
SpecTreatable patients
vey
0 6 12 18 24 30 36 42 48
Time (Month)
‘Treatments : Runt
‘SpecT reatable Patients : Runt
‘SpecT reatable Patients with Diagnostic : Run. 3333 3 3
Figure 23: Specialist Physician Patient Stocks
Time Spent Per Examinations and Tests
9 Pe I
6
0 6 12 18 24 30 36 42 48
Time (Month)
SpecTSPE : Run1
SuTSPE : Runt
TimeSpentPerTests : Runl. —3—~3—3—_3—_3—_ 3—_ 3
Figure 24: Time Spent Per Examination and Tests
GoalSpecRevenue
0 6 12 18 30 360 42—— 48
24
Time (Month)
:Runl
Figure 25: Specialist Physician Goal Revenue
6.2. Economic Crisis-Budget Cuts
In this scenario, government faces a big economic crisis. MOH cannot provide high
incentives for medical activities anymore. Due to decreases in performance payment system,
physicians prefer working in private sector. As a result, public health employees are lower
than that in base model. Moreover, private hospitals increase the physicians’ base revenue to
increase their productivity and market share.
Since the performance revenue of physicians is far lower than that in the base run,
physician’s revenue pressure is expected to be high. Moreover, decreases in health resources
do not solve the unmet health demand problem. As a result, examination and surgery
crowding increases due to inadequate number of health employees.
In Figure 26, it can be seen that times spent per examination and test decrease due to
increases in crowding and revenue concern of hospital and physician. Physicians can increase
their revenue by improving their productivity.
Time Spent Per Examinations and Tests
10
¢ See CEE EEE Et
:
4
0 6 12 18 24 30 36 42 48
Time (Month)
SpecTSPE : Run4
SurTSPE : Run4
Ti erTests : Run
Figure 26: Time Spent Per examinations and tests
Doctor Revenue
4,000
TL/(Month*physicians)
x
a
Ss
Ss
isn
0 6 122 18 24 30 36 42 48
Time (Month)
1,000
SpecRevPerMonth : Runt
SurRevPerMonth : Run4
DiagPhyRevPerMonth : Run4. 2——3—3—_3—_ 3-—_3—_-
Figure 27: Physician Revenues
Quality Output- 1
26
Se oH 4
0 6 12 18 #24 30 36 42 48
Time (Month)
urCorDiagRatio : Run4d
SpecC io : Run4d
SurCorTreatRatio : Run4
SpecCorTreatRatio : Run4
Figure 28: Correct Treatment and Diagnose Ratios
6.3. No PBPS at all
If payment system is not based on medical performance of physicians, they still have
revenue pressure but they don’t have opportunity to improve their performance for increasing
their revenue.
In this scenario, the effects of PBPS are excluded from model. Physicians’
productivity is only affected by the hospital crowding. Due to the hospital crowding,
physicians may spend less time per examinations and medical activities.
Although physicians’ revenue is lower than their goal revenue, they can’t increase
their revenue by performing more medical activities. These behaviors can be seen in Figure
29 and Figure 30.
No significant dynamics are observed in this scenario. Absence of revenue-related
effects eliminates the adverse effects of PBPS on quality indicators. This can be seen in
Figure 31.
Time Spent Per Examinations and Tests
ner
0 3 6 9 12 15 18 21 24 27 30 33 36 39 42 45 48
Time (Month)
SpecTSPE : NOPBPS Effect —1s——a—2—3_3 334 3
Surg SPE : NOPBPS Effect ——_2—-2—2—_2 222 2 —
TimeSpé Tests : NOPBPS Effect
Figure 29: Time Spent Per Examination and Test
Doctor Revenue
6,000
o
=)
=
Ss
w
=
=
S
r
‘TL/(Month*physicians)
S
=
Ss
Ss
2,000
0 6 12 18 24 30 36 42 48
Time (Month)
SpecRevPerMonth : NOPBPS Effect. +2122 3.
SurgRevPerMonth : NOPBPS Effect —2—2——_2—_2—_2—_2—_
DiagPhysRevPerMonth : NOPBPS Effect. ——3—3—3—_3—3—
Figure 30: Physicians’ Revenues
Quality Output-1
0.4
0 6 12 18 24 30 36 42 48
Time (Month)
SuryConDiagRatio : NOPBPS Effect. ——3——3——4——3 3
SpecComDiagRatio : NOPBPS Effect. —-——2——2_»_2_> —
SuryConTreatRatio : NOPBPS Effect. —3——3——3 3 33
SpecCon'TreatRatio : NOPBPS Effect. —+——+——+4++
Figure 31: Correct diagnostic and treatment ratios
6.4. Abundant D: d-Inad Specialist Physicians
In this scenario, number of physicians is decreased from 20 to 10. Hospital
examination crowding increases as expected. Physicians experience the pressures of revenue
and crowding. As a result, they try to increase their productivity by decreasing TSPE. This
behavior can be seen in Figure 32.
SpecTSPE
minutes/people
3 ©
0 6 12 #18 %24 30 36 42 48
Time (Month)
SpecTSPE : LowPhysicians ——s—-z—3— 44
Figure 32: Time Spent Per Examination
By performing more medical activities, physicians increase their revenue, which can
be seen in Figure 33.
SpecRevPerMonth
2,500
TL (Month physicians)
oa
S
s
s
0 6 12 18 24 30 36 42 48
Time (Month)
SpecRevPerMonth : LowPhysicians ———+———_—_+—.
Figure 33: Specialist Physicians Revenue per Month
6.5. Inadequate Demand-L ow Performance Payments
In this scenario, there is inadequate demand for examination. Potential external
examination demand to physicians is decreased to 3000 people/month. In addition to this,
performance point per examination is decreased to 10 points/examination.
Although, specialist physicians’ revenue is far lower than their goal revenue, they do
not have opportunity to increase their income by examining more patients. The reason behind
this situation is inadequate examination demand. Thus, time spent per examination doesn’t
decrease as a result of specialist’s revenue concern, as expected. This behavior can be seen in
Figure 34.
SpecT SPE
minutaypronte:
0 6 12 18 24 30 36 42 48
Time (Month)
SpeeTSPE
Figure 34: Time Spent Per Examination
DISCUSSION AND CONCLUSION
The aim of this study is to investigate dynamic impacts of performance based payment
system (PBPS) on health service outputs. PBPS implementation in Turkey considers public
health centers as revenue generating places. In order to meet health demand and increase
medical productivity, PBPS has been active in second step public hospitals since 2004.
Considering the long implementation history and share in medical operations, second step
public hospital is selected and a model that represents the dynamic effects of PBPS on these
hospitals and physicians is built.
Physicians’ revenue and their response to government policies are related. With PBPS,
physicians have a chance to improve their living standards by obtaining more performance
revenue. If physicians already ear satisfactory salaries, then quality variables are expected to
be positive with PBPS. In the base run, time spent per examination, performance points for
medical activities, health resources and external demand are seen as main factors affecting the
system behavior. According to simulation runs, there is a negative relationship with
physician’s revenue concem and health service quality, because of the fact that physician’s
revenue is strongly based on his/her productivity.
In scenario analysis, when physicians’ revenue concem is high, physicians tend to
spend less time per medical activity (examination, diagnostic and treatment) in order to
increase their revenue. Quality indicators decrease as can be predicted. Inadequate treatment
stocks increase and reach relatively high equilibrium values in 30 months. In another
scenario, government decides to decrease the health expenditures and cut down performance
points per medical activities. However, there is abundant demand for medical service and
physicians can increase their productivity to increase their revenues. Therefore, inadequate
and low quality treatments result, as well as crowding in hospitals. Efforts to decrease health
expenditures end in failure because of the very structure of payment system.
To sum, this study is an initial effort for understanding dynamic effects of PBPS and
presents base model for further studies. As further research, the relationships and competition
between public and private health sectors can be explicitly modeled investigated. Moreover, a
university hospital model may be built for investigating different impacts of PBPS. Thus, the
effects of hospital revenue on educational and research activities may also be investigated.
REFERENCES
[1] | CRS Report for Congress “Life Expectancy in the United States” 16.08.2006.
[2] CMS (Center for Medicare and Medicaid Services). 2007. National Health
Expenditure Data. http://www.cms.hhs.gov/NationalHealthExpendD ata/02[24 January 2007].
[3] Jack Homer, Gary Hirsch and Bobby Milstein, 2007, “Chronic illness in a complex
health economy the perils and promises of downstream and upstream reforms” System
Dynamics Review Vol. 23, pp. 313-343.
[4] OECD Health Statistics http://www.oecd.org.
[5] Marc Pomp, 2010, “Pay for Performance and health outcomes: a next step in Dutch
health care reform” Background paper for the Council for Public Health and Healthcare.
[6] Marko Vujicic, Susan Sparkes, Salih Mollahaliloglu “Health Workforce Policy in
Turkey Recent Reforms and Issues for the Future” July 2009.
[7] “Turkish Health System Evaluation” OECD Health Report 2008.
[8] Performance Management in Healthcare: Ministry of Health “Performance Based
Payment System Report“Ankara-2007.
[9] Turkey Ministry of Health: Health Service Report 2010.
[10] Ali Gazi “Analysis of Effects of Performance Based Additional Payment System on
Patients and Health Employee “Gazi University Social Science Institute A nkara-2006.
{11] Turkey Ministry of Health: Turkey Health System Classification 2003
[12] WHO: A Process for Making Strategic Choices in Health Systems 2006
[13] Tubitak:’Public Hospital Contributions to Number of Academic Publication and
Citations (1981-2006)” 2008
[14] Tubitak: “Scientific Publications & Citations Performance of Turkish Universities
(1981-2007) 2009
[15] Sterman, J.D. Business Dynamics: Systems Thinking and Modeling in a Complex
World. McGraw-Hill, Boston, 2000.
[16] Barlas, Y 1996. Formal Aspects of Model Validity and Validation in System
Dynamics, System Dynamics Review, Vol.12, No.3, pp.183-210
[17] Ann Van Ackere, Peter C. Smith “Towards a macro model of National Health Service
waiting list” System Dynamics Review Vol.15 No.3 Fall 1999
[18] Eric Wolstenholme, 1999, “A patient flow perspective of U.K. Health Services:
Exploring the case for new “‘intermediate care" initiatives” System Dynamics Review. Vol.
15, 253-271.
(19] Turkish Physician Union, 2009, “PBPS with evaluation of physicians” Ankara.
[20] Cahit Korku “The effects of PBPS on health service quality: Evaluation of Hospital
Managers and Health Employees” Hacettepe University Health Sciences, Ms. Thesis 2010.
