Examination Of The Technical Efficiencies Of Family Health Centers In Turkey
Abstract
In this study, it was aimed to determine the potential improvement areas of the family health
centers (FMU) operating in Turkey, the efficiency scores, the input uses that lead to FHC's
being efficient or inefficient, and the potential improvement areas that need to be improved in
order to be sufficient for FHC's being inefficient. For this purpose, a total of 6.902 FHC, serving
throughout Turkey in 2015, was taken as a decision-making unit (DMU) and analyzed by
dividing it into groups determined by the Ministry of Health. In this approach, four input and eight
output variables were used. The method was developed by Charnes, Cooper, and Rhodes
(CCR) in 1978, and the multiplication method of data envelopment analysis (DEA) was used.
Since it is aimed to determine the potential improvements on input and to parse decisionmaking units better, the DEA model was set up under the assumption of constant return to scale
based on the input-oriented scale. According to the results, the mean efficiency values of FHC
groups were between 53.00% and 71.80% and the efficiency scores were found to be
statistically significant. The number of persons enrolled in family physician, the number of family
physician units and the number of family physicians were found to be the main variables
affecting the efficient or inefficient functioning of the units. According to the results of the study,
to increase the efficiency of FHCs, improvement should be made on persons registered with
family physician first. Then, it is recommended to improve the number of family physician units
and the number of Family Physicians.
Keywords: Family Health Center, Family Physician, Performance Evaluation, Efficiency, Data
Envelopment Analysis.
Introduction and Background
Several models are applied to provide primary health care services throughout the
world. One of these methods, family physician (FM) system, especially since the
second half of the twentieth century has come to the fore in many nations. In line with
the world trend, the FM system which is intended to be applied in many years in
Turkey's health system and which is among the public policies has not been applied for
various reasons until the 2000s (Algın et. al., 2004, s.255).
With the project “Health Transformation Program” (SDP) which was financed by the
World Bank in 2003, the implementation of the FM system has been on the agenda in
Turkey. In this context, the FM model was introduced as a guide in some provinces in
2005 in Turkey, and it started to be applied in all country as of 2010 (Akdağ, 2011,
s.25). In the new system, family health centers (FHC) consisting of family physician unit
(AHB) and family physician unit consisting of at least one family physician were
established in order to carry out FM services.
With the SDP project, there have been significant changes in management philosophy
as well as the administrative arrangements related to the FM system. Within the scope
of these modifications, the effective, efficient and fair presentation of health services
has been identified as a fundamental component of SDP. Also, a performance system
has been developed to monitor the production of managers and employees efficiently.
In other words, individual and institutional performance evaluation in primary health
care has become even more prominent.
In performance measurement, it is possible to obtain information about the
performance by measuring the various dimensions that form the performance. The
efficiency, which is one of these dimensions, can be defined as producing as much
output as possible with the resources that are generally owned or producing outputs
with minimum resources (Porcelli, 2009, s.3). Although some methods are used in the
measurement of efficiencies, data envelopment analysis (DEA) is commonly used to
evaluate the efficiencies (Wang and Lan, 2013, s.182).
It is stated that DEA provides essential information in terms of managerial decision
making (Cooper et.al, 2011, s.2). When DEA evaluated in terms of the health
sector, it can help health managers to evaluate the relative performance of health
institutions, to determine the necessary performance in the health sector, and to
determine the fundamental ways to improve the performance of the units (Ozcan,
2008, s.17).
Due to the application profits mentioned, there are some studies in Turkey where
primary health care is evaluated with the DEA (Such as, Kayalı et.al., 2004; Üner,
2006; Uyar, 2009; Özata and Sevinç, 2010; Bircan, 2011). These studies were carried
out in Turkey when primary health care was carried out according to the health care
system. In the primary health system, it is essential to provide health services for both
individuals and the environment, and in the FM system, to provide health services for
individuals and families. Therefore, in this study, it can be stated that the efficiency of
different service units is evaluated and the work in this regard differs from the work
done in the primary health care system. In a study conveyed during the FM system
period (Erinç, 2013), the efficiency of family physicians working in a province was
evaluated. In our study, the institutional performance of FHC services across the
country was evaluated within the framework of the data accessible. Due to the
differences in the duty descriptions of both the application area and the units evaluated,
this study is separated from the previous studies.
Factors such as that the FM system is an innovative unit in the Turkish health system
than other systems, that data on the production of the service are available, that
primary health care services are less focused on the efficiency evaluation, that DEA is
widely available in the health sector, that no studies dealing with the efficiency of FHC,
and that the results obtained can be used in practice have a motivational impact on the
conduct of this study.
This study aims to determine the technical efficiency of FHC in Turkey to provide FM
services, to determine the factors that lead to FHC efficiency/inefficiency in FHC, and
to develop scientific-based application recommendations to improve the efficiency of
these services offered by the public.
MATERIALS AND METHODS
In this study, DEA was used to evaluate technical efficiency. DEA can be used in public
areas, and non-profit units (Huang and McLaughlin, 1989, p.144), widespread use in
the health sector, a large number of DMU and multiple inputs and output variables
belonging to these units will be evaluated and the efficiency of the variables used in the
functional association is not needed (Pelone et al.(2015), due to features such as the
ability to determine the recommendations to be useful for units identified as inefficient,
DEA has been decided to be used as an analysis. The measures taken within the
scope of the DEA are as follows:
●
Selection Of Decision-Making Units
In Turkey, 6.902 FHC served in 2015 (Köse et al., 2016). In DEA, since DMUs must be
comparable to each other and serve in similar environmental conditions, the grouping
of FHCs in force in the current practice has been taken into consideration, and it has
been ensured that units are analyzed separately according to the groups to which they
belong. In this process, although they are correlated to the same FHC, it has been
observed that some of the family physician units are in different groups. In this case,
the FMU's were considered as a separate FHC, and the number of FHC in the analysis
was increased from 690.2 to 7.400, so that grouping could be done correctly. FHCs,
which may reduce homogeneity from the said 7,400 FHCs are excluded from the scope
of analysis (for example, FHCs in prisons where pregnancy and baby-related services
are not available). Also, it has been observed that the values of some DMU are not in
the obtained data set. As a result of the elimination of 255 FHC from the analysis
scope, 7.145 FHC was determined as a decision-making unit. These FHCs were
divided into groups and analyzed separately. The number of DMU in each group is as
follows:
577 FHC in group A,
911 FHC in group B,
507 FHC in group C,
2.017 FHC in group D,
2,133 FHC in group E.
●
Determination of Input and Output Variables
The first state of the input and output variables planned to be used in this study was
determined by scanning the efficiencies carried out by the DEA in primary health care
services. The parameters were revised considering the characteristics of the FM
system applied in Turkey, the usage of input and output variables in all DMUs and the
availability of the data. Regarding the revised variables, considering the opinions of the
personnel working in FM units, the provincial managers in performance monitoring and
evaluation, the head of department personnel in monitoring and evaluation, and the
managers, it has been decided that the variables will become ultimate. The names,
definitions, and reasons for the variables used in the study are given in Table 1.
Table 1. Output Variables Used In The Study
The Name of the
Variable
Definition
Justification For Use
●
Total
number
of
medical examinations
Total Number Of
●
carried out by FHC
Medical
within one year
Examination
O
u
t
p
u
t
s
●
The total number of
screening
for
Total number of
colorectal, cervical and
Cancer Screening
breast cancer in a year
by FHC
Number
puerperant
monitoring
Number
pregnant
of Total
number
of ●
puerperant monitored
by FHC within one year
of Total
number
of
pregnant monitored by
A variable used in all studies in
this field
Being one of the basic services
expected
from
family
physicians (Official Journal
January 25, 2013, a.4; Official
Journal, May 25, 2010, a.4).
Being one of the main
preventive
and
protective
services for the expected from
family
physicians
(Official
Journal, January 25, 2013, a.4;
Official Journal, May 25, 2010,
a.4).
All of these variables are the
essential services expected
from family physicians (Official
Journal, 25 January 2013, m.4;
Official Journal, 25 May 2010,
monitoring
FHC within one year
Total
number
of
Number
of
children monitored(1-5 ●
children
years) by FHC within
monitoring
one year
Total number of infant
Number of infant (0-12
months)
monitoring
monitored by FHC
within one year
●
Number
of
Total number of MMR
measles, mumps,
vaccines administered
rubella
(MMR)
by FHC within one year
vaccine done
Total
number
of
combination
vaccine
[diphtheria, pertussis,
Number
of tetanus,
poliomyelitis
combination
and
haemophilus
vaccine done
influenzae
type
B
(meningitis)]
administered by FHC in
one year
m.4).
These variables have criteria
for determining the monthly
performance coefficients of
Family Physicians (Ministry of
Health, 2007; World Bank,
2013).
For these variables, if the
performance
objective
indicator level is below 90%, a
warning score is given to
family physicians (World Bank,
2013).
Tablo 2. Çalışmada Kullanılan Girdi Değişkenleri
The Name of the
Variable
Definition
Justification For Use
Number
of
family ●
physicians in FHC
The number of Number
of
family ●
family
health health personnel in
personnel
FHC
●
The number of Number of FM units
FHC units
connected to FHC
The number of FM
I
n
p
u
t
s
●
Population
registered to FHC
Mid-year
population
registered with FHC
●
●
Be the primary input variables
used in all studies in this field
Being the two essential titles
defined in the law to provide
FM service
Demand to determine whether
a modification is necessary for
the number of units connected
to FHC
Identification of payments to
family physician personnel
according to the registered
population (Official Journal,
April 16, 2015)
The primary determinant of
the
demand
for
family
physician services
Data Acquisition and Classification
In order to obtain the data, a written application was made to the Turkish Public Health
Institution, and the General Directorate of Health Information Systems of the Ministry of
Health and legal permission was obtained. The data obtained at the FM unit level are
combined with the groups of FHC and FHC to which they are affiliated.
●
Determination Of Model
For the first time, Charnes, Cooper, and Rhodes (1978) suggested the DEA method
and stated that the efficiency of any DMU could be obtained by maximizing the
weighted output of the unit and the ratio of the weighted input used by the unit to the
maximum output of the unit. For the mathematical representation of this expression,
assume that DMU evaluates n, that DMU uses m number of inputs and produces s
number of outputs. j; DMU, i; input of DMU xij, r; output of DMU yrj, shown as. In light
of this information, the mathematical representation of the CCR model DEA is as
follows (Charnes et al., 1978, p.430);
Objective Function:
∑𝑠𝑟=1 𝑢𝑟 𝑦𝑟𝑜
ℎ𝑜 =
∑𝑚
𝑣𝑖 𝑥𝑖𝑜
𝑖=1
(1)
Limitations:
∑𝑠𝑟=1 𝑢𝑟 𝑦𝑟𝑗
≤ 1
∑𝑚
𝑣𝑖 𝑥𝑖𝑗
𝑖=1
𝑣𝑖 ≥ 0
𝑢𝑟 ≥ 0
𝑖 = 1,2 … . . , 𝑚
𝑟 = 1,2 … . . , 𝑠
𝑗 = 1,2 … . . , 𝑛
is the weighted output sum,
is the weighted input sum. Max. ho, is the
objective function that determines optimal input-output weights and maximizes the
value. The first limitation shows that the ratio of the weighted output totals in all DEAs
to the weighted input totals is equal to 1 or less than 1. This limitation has been
introduced so that the efficiency limit can be within a certain range and the upper limit
can be set to 1 (Agriculture, 2001, p.50). In other limitations, input weights (vi) and
output weights (ur) are determined to be greater than or equal to zero. It is stated that
the result of the solution under these limitations, if the ho value is 1 o; the unit is
efficient, and if the unit is less than 1 o; the unit is inefficient (Charnes et al., 1978,
p.430).
Researchers must decide on some points when applying the DEA. The first is that
since the efficiency measurements in DEA can be measured in input and output
direction, the researcher should determine whether to focus on reducing inputs or
increasing outputs. Second, the researcher should determine the scaling hypothesis
because measurements can be performed under constant return to scales or variable
returns according to scale in the DEA. Finally, the fractional expression in Equation (1)
can be arranged according to the linear programming technique, making it easier to
produce solutions. In linear programming, two solutions are closely related to each
other. The first one is called primal; the second one is called dual. In the literature of
the DEA, it is generally referred to as "envelopment" to primal models, and "multiplier"
to dual models. Although envelopment and multiplier models at the DEA provide the
same efficiency results, each solution method offers additional information in different
details as well as efficiency results. In the DEA, the envelopment solution method can
be used for benchmarking and targeting, while the multiplier solution method can be
used for determining the strengths and weaknesses of DMUs, thus determining the
intervention fields for potential improvement. Therefore, the researcher should
acknowledge that the research should determine a model that is appropriate for the
purpose.
In the studies aimed at policy regulation, it is stated that the assumption of constant
return to the scale would be more appropriate. Also, it is expressed by some authors
that if the input and output variables have more control over which direction it would be
appropriate to determine (Pelone et.al, 2015, p.6). Besides, Ozcan (2008, p.23) is
recommended that input-oriented models should be implemented because the
dominance over inputs in health services is higher than the dominance over outputs.
The studies in primary health care area mostly support the affirmed statements which
show that analyses are carried out under the assumption of input-oriented and constant
returns based on the scale.
In this study, it was decided to use the multiplier DEA solution method as it focuses on
developing application recommendations in order to increase efficiency in the FM
system applied in Turkey. Also, since the dominance over inputs is higher than the
output, it is decided to use the input orientation and the constant return to scale
hypothesis on the scale to differentiate the efficiency along the units better.
The equation of the method practiced is as follows:
Objective Function:
𝑠
𝜂𝑘 = ∑
𝜇𝑟 𝑦𝑟𝑘
(2)
𝑟=1
Limitations:
𝑚
∑
𝜔𝑖 𝑥𝑖𝑘 = 1
𝑖=1
𝑠
𝜇𝑟 , 𝜔𝑖 ≥ 0
𝑚
∑
𝜇𝑟 𝑦𝑟𝑗 − ∑
𝑟=1
𝑖=1
𝜔𝑖 𝑥𝑖𝑗 ≤ 0
( j=1,….n) (r=1,….s) (i=1,….m)
Here;
𝜂𝑘 :
The objective function,
𝜇𝑟 :
k. for DEA r. weight of output,
𝜔𝑖 :
k. for DEA i. weight of input,
𝑦𝑟𝑘 :
k. production of DEA r. output,
𝑥𝑖𝑘 :
k. used by DEA i. input,
𝑦𝑟𝑗 :
j. DEA produced r. output,
𝑥𝑖𝑗 :
j. DEA used i. input
●
Analysis Of The Data
In order to determine the technical efficiency of FHC, Gurobi 7.0.1, one of the widely
used solvents for optimization problems, was used. This mathematical programming
solvent was coded and analyzed in the Python programming language. Statistical
analysis was performed using the Statistical Package of Social Sciences (SPSS) to
determine whether efficiency scores differ significantly between groups and fields.
Thematic maps created to show the efficiency values were created in ArcGIS 10.1
program.
RESULTS
7.145 FHC, which was included in the study, was analyzed separately according to the
groups they were affiliated with. The findings of the efficiency scores of the analysis
results are given in Table 3.
Table 3. Technical Efficiency Findings By Groups Of FHC
Group
Efficient
Inefficient
f
f
%
%
Mean
Efficiency
SD
Min
Max
A (n=1577)
76 4,82% 1.501 95,18%
66,05%
16,22%
9,36%
100%
B (n= 911)
40 4,39%
871
95,61%
66,65%
15,07%
22,77%
100%
C (n= 507)
27 5,33%
480
94,67%
71,80%
15,22%
33,65%
100%
D (n= 2017)
61 3,02% 1.956 96,98%
62,83%
15,52%
14,14%
100%
E (n= 2133)
63 2,95% 2.072 97,05%
53,00%
18,25%
2,89%
100%
f: frequency, %: percent, h: standard deviation, Min: minimum, Max: Maximum
Note: Two digits after the comma are taken into account in terms of percentage values' ease of
explanation and interpretation.
The most efficient unit is in group A, proportionally in the Group C. Group C is also the
group with the highest mean efficiency value. Group E with the most FHC is the group
with the lowest mean efficiency score, and the group with the lowest efficient unit
proportionately. In the Group D and E with the most FHC, the mean score of Group D
is about 10% higher than the mean score of Group E, although there is approximately
the same percentage (3%) of the efficient units. The highest variation in the efficiency
scores was observed in Group E, and the lowest variation was observed in Group C
(table 3).
It was demanded to test whether there was a significant difference between the
efficiency values of the groups. For this purpose, firstly, the Kolmogorov-Smirnov test
was used to test whether the efficiency scores were average. As a result of the
normality test, efficiency scores do not show a normal distribution (p=0,0001; p<0,05).
Table 4. Normality Test of The Group Efficiency Scores
Test
Statistic
Efficiency Score
0,028
Degree Level Of
Of
Significan
Freedom
ce*
7145
0,0001
* p= 0,05 as taken.
Because the efficiency scores did not show a normal distribution of efficiency values
whether there is a significant difference along groups has been tested with the help of
the Kruskal-Wallis test which is a nonparametric test. As can be seen from Table 5,
efficiency values showed a significant difference along groups (χ2= 903,333; p= 0,0001;
p<0,05).
Table 5. Significance Test Of Efficiency Scores By Groups
X-square
Degree of Freedom
Level of Significance*
* p= 0,05 as taken.
Efficiency Score
903,333
4
0,0001
The Dunn-Bonferroni posthoc test was used to determine which groups were included
in this significant difference. The findings are displayed in Table 6. As a result of the
analysis, it was found that the difference between Group A and Group B was not
statistically significant. On the other hand, all binary difference along groups (except
group A and B) was found to be statistically significant for group comparisons (Table
6). Table 3 shows that group A and group B mean efficiency values are very close to
each other, the highest mean efficiency is in Group C and the lowest mean efficiency is
in Group E. In particular, since the mean efficiency of Group E is significantly lower
than the mean efficiency of other groups, it can be said that group E is the group that
makes the difference along groups.
Table 6. Paired Comparison Of The Efficiency Values Of Groups
Level of
Groups
Test Statistic Significance (p)
Group A - Group B
-1,184
0,236
Group A - Group C
-6,637
0,0001*
Group B - Group C
-5,226
0,0001*
Group D - Group A
5,482
0,0001*
Group D - Group B
5,851
0,0001*
Group D - Group C
10,529
0,0001*
Group E - Group A
22,661
0,0001*
Group E - Group B
20,260
0,0001*
Group E - Group C
22,090
0,0001*
Group E - Group D
18,298
0,0001*
p= 0,05 as taken.
* The difference along the groups is statistically significant.
The correlation coefficients of the groups were tested with Spearman's Rho correlation,
and the outcomes were displayed in Table 7. According to the correlation table;
between Group C with Group A, Group C, and Group D and between Group B and
between Group C with Group E, it has been found that there is a statistically significant,
positive, and relatively weak relationship. There was also a statistically significant,
relatively weak and negative correlation between Group E and Group D. There was no
statistically significant difference along the other groups (Table 7).
Table 7. Correlation Coefficients Along Groups According To Efficiency Values
Group A
Group B
Group C
Group A
1
Group B
-0,045
1
Group C
0,096*
-0,065
1
Group D
0,027
-0,013
0,133**
Group E
0,004
0,096**
0,092*
* p<0,05 was significant, ** p<0,01 was significant.
Group D
Group E
1
-0,067**
1
Input weights obtained through multiplier DEA solution method were used to determine
efficiency scores in order to determine the strengths of DMUs and potential
improvement fields of DMU (Cooper et al., 2006). The strengths of the efficient DMU
are calculated as follows: As a result of the analysis, the calculated input weights and
real values are multiplied by the virtual input (Cooper et al. 2006, p.21) usage
performances of inputs have been obtained. The highest value input or inputs are
interpreted as the strengths of the units by taking into account the performance values
of each input. While the values were considerably high on a given input, it was
observed that several inputs were very close to each other in some units. By taking into
consideration, the highest values (5% or less) that are very close to each other, input
performances that enable all units to perform efficiently were determined. Potential
improvement areas of inefficient units were determined by the same method. The only
difference here is that these weights are interpreted. In other words, the inputs of the
units in which it operates are the potential improvement fields where the usage
performance values are highest, which should be interfered by the units in order to
increase the efficiency. Table 8 shows the performance of utilizing inputs that lead to
the efficiency of DMUs performing efficiently. In Table 9, potential improvement fields
that should be focused on to increase efficiency levels of FHCs which are found to be
inefficient are given.
Table 8. Input Usage Performances (Strengths) That Lead To Efficient FHCs
Groups
Number of
FM
Number of
Family
Physician
Number of
Number of FMU
Persons
Registered to the
Personnel
FHC (AHK)
A (n=76)
25,64%
5,13%
11,54%
57,69%
B (n=40)
15,56%
4,44%
22,22%
57,78%
C (n=27)
55,17%
0,00%
17,24%
27,59%
D (n=61)
11,11%
9,52%
31,75%
47,62%
E (n=63)
3,03%
4,55%
50,00%
42,42%
As shown in Table 8, the best performance of the units working efficiently in groups A,
B and D is the success of determining the number of AHK. The majority of the units
that are efficient in Group C have shown the best performance on the FM number
variable. It was found that the number of family health personnel did not significantly
affect any of the FHC in Group C. In the efficient units in Group E, the number of FMU
and the number of AHK variables are the most fundamental variables in the efficient
output of the units. The variable that had the least impact on the performance of the
efficient units in all groups except Group E was found to be the number of family health
personnel.
Table 9. Input Performance That Causes Inefficient FHC To Be Inefficient
(Potential Improvement Fields)
Number
Number of
Groups
FM
of
Family Health
Personnel
Number of FMU
Number of AHK
A (n=1501)
14,36%
0,93%
14,48%
70,23%
B (n=871)
2,30%
0,66%
13,13%
83,92%
C (n=480)
41,33%
0,00%
4,23%
54,44%
D (n=1956)
0,54%
1,61%
38,34%
59,51%
E (n=2133)
1,59%
2,91%
30,18%
65,33%
In Table 9, input usage performances are given for FHCs that are inefficient. If the units
in the groups have taken the highest value on which variable, these variables are the
base fields that need to be improved by the relevant units. In Group A, 70.23% of
inefficient units were determined as inefficient because of their excessive number of
AHK, 14,48% number of FMU, 14,36% number of FM, 0,96% number of family health
personnel due to their performance on variables were determined inefficient. In Group
B, a significant number of inefficient units (83.92%) were more than the number of
AHK, and 13.13% were inefficient due to the number of FMU variables. In Group B, the
variables of the number of FM and family health personnel are very weak in influence
on the increase in the efficiencies. The units that are inefficient in Group C are
inefficient because of the variables 54.44%, the number of AHK, 41.33%, the number
of FM, 4.23%, and the number of FMU. The number of family health personnel was
determined to have no impact on the efficiency of FHC in Group C and found to be
ineffective. In Group D, 59.51% of the inefficient units were found to be inefficient due
to the excess number of AHK and 38.34% due to the number of FMU. On the other
hand, a deficient proportion, FM number and Family Health number variables (0.54%
and 1.61%, respectively) were inefficient. In Group E, 65.33% of the units found to be
inefficient because of their excess number of AHK, 30.18% of FMU, 2.91% of family
health personnel and, finally, 1.59% of FM number variables.
Since it is thought that the findings outlined above will provide a more detailed
assessment of the regional and administrative decisions in this direction, it is
considered appropriate that the efficiency scores should be given according to the
Level 1 of the Nomenclature of Units for Territorial Statistics (NUTS).
Evaluation of Efficiency Scores by NUTS-I Regions
Group A
In Table 10, the NUTS-1 level of the FHCs in Group A is based on the technical
efficiency status, the distribution of efficiency scores, and which inputs FHCs should
focus on in order to be able to perform efficiently.
Table 10. Efficiency Status of FHCs in Group A according to NUTS-1 Level
Level 1
Regions
İstanbul
West Marmara
East Marmara
Aegean
Mediterranean
West Anatolian
Central
Anatolian
West Blacksea
East Blacksea
North East A.
Middle East A.
South East A.
Mean
Effici Ineffici
Number
efficienc
ent
ent
of FM (%)
y (%)
Number of
Family
Health
Number
Personnel of FMU
Number
of AHK
-
-
55,68
67,45
65,58
69,19
68,31
64,06
-
1
2
2
3
1
0
-
-
5
166
61,22
12
1
10
149
3
1
2
0
34
-
58,48
60,33
69,85
60,17
76,97
-
1
2
0
1
2
-
-
According to this, among the NUTS-1 levels of FHC in Group A, the regions with the
most efficient units are the Southeast Anatolia Region (34), Aegean Region (16) and
Mediterranean Region (8) respectively. In Group A, none of the 173 FHCs in Western
Anatolia and Middle Eastern Anatolia regions were found to be efficient. It was found
that the region with the highest efficiency mean was the Southeastern Anatolia region
(76.97% ± 14.36%), and the region with the lowest was Istanbul region (55.68% ±
12.91%). In order for FHCs in all regions of this group to work effectively, it is
necessary to reorganize them by focusing primarily on the number of FHCs inputs.
Secondary, almost the same importance in many regions and improvements should be
made primarily on the number of FM and FMU. In this group, the number of family
health personnel in all regions was determined as the least significant variable. The
Thematic Maps of NUTS-1 level, which is formed according to the mean efficiency
values of FHC in Group A, is displayed in map 1.
Map 1. Thematic mapping of Group A FHCs according to NUTS-1 Level
Group B
Table 11 shows the technical efficiency situation of FHCs in Group B at NUTS-1 level
and which inputs should be focused on in order for FHCs to function efficiently. The
most efficient unit in Group B was found in Southeastern Anatolia (10), Mediterranean
(9) and Aegean (8) regions. In this group, none of the FHCs performing in Central
Anatolia, Eastern Black Sea, Northeastern Anatolia, and Middle Eastern Anatolia
regions were efficient.
Table 11. Efficiency Status of FHCs in Group B according to NUTS-1 Level
Level 1 Regions
İstanbul
West Marmara
East Marmara
Aegean
Mediterranean
West Anatolian
Central Anatolian
West Blacksea
East Blacksea
North East A.
Middle East A.
South East A.
Efficient
Inefficient
-
-
Numbe
r of
Mean
Family
Efficien
Health Numbe Numbe
cy (%)
Numbe Person
r of
r of
r of FM
nel
FMU
AHK
57,73
71,29
70,93
68,11
72,95
62,34
56,93
63,18
60,49
66,87
61,34
78,47
-
-
-
-
In Group B, the highest mean efficiency scores were observed in the regions of
Southeastern Anatolia (78.47%), Mediterranean (72.95%) and Western Marmara
(71.29%), while the lowest mean efficiency scores were observed in the regions of
central Anatolia (56.93%), Istanbul (57.43%) and Eastern Black Sea (60.49%). In order
for FHCs in Group B to be efficient, the AHK number input appears to be the only field
in most regions to be almost regulated, while the FMU number input in the West
regions appears to be considered as the priority field to be regulated. Improvements in
the number of FM and family health care personnel are relatively low in the potential
contribution of the FHCs in this group to the efficient performance. The Thematic Maps
of NUTS-1 level, which is formed according to the mean efficiency values of FHC in
Group B, is placed in Map 2.
Map 2. Thematic mapping of FHCs in Group B according to NUTS-1 Level
Group C
The intervention areas of FHCs in Group C are given in Table 12, which may increase
the technical efficiency and efficiency of the FHCs according to NUTS-1 level. The
regions with the most efficient unit in Group C are Southeastern Anatolia (8), Aegean
(6) and Mediterranean (3) regions. In Central Anatolia and Western Black Sea regions,
no units are working efficiently.
Table 12. Efficiency Status according to NUTS-1 Level of FHCs in Group C
Level 1 Regions
İstanbul
West Marmara
East Marmara
Aegean
Mediterranean
West Anatolian
Central Anatolian
West Blacksea
East Blacksea
North East A.
Middle East A.
South East A.
Numbe
r of
Mean
Family
Efficien
Inefficient Efficiency
t
Health
Numbe
(%)
Numbe Person Number
r of
r of FM
nel
of FMU AHK-
-
60,74
73,02
78,43
79,65
76,40
74,43
64,32
61,51
69,97
72,17
69,30
81,25
-
-
-
-
In Group C, the highest mean efficiency scores were found in the regions of
Southeastern Anatolia (81.25%), Aegean (79.65%) and Eastern Marmara (78.43%),
while the lowest mean efficiency scores were observed in the regions of Istanbul
(60.74%), western Black Sea (61.51%) and Central Anatolia (64.32%). In this group,
amendments should be made mainly on FM and AHK number variables to increase
efficiency. In all regions in this group, the number of family health personnel input may
not be used as a tool to increase the efficiency of FHCs. In Map 3, the Thematic Map of
NUTS-1 level created according to the mean efficiency values of FHCs in Group C is
given.
Map 3. Thematic mapping of Group C FHCs according to NUTS-1 Level
Group D
In Table 13, FHCs in Group D there are; technical efficiency status, distribution of
efficiency scores, and potential improvement areas for FHCs working without efficiently
to be able to work efficiently. The regions with the most efficient units are
Mediterranean (12), Southeastern Anatolia (12) and Western Marmara (10) regions.
The least efficient unit is in the Eastern Black Sea (0), Middle East Anatolia (1),
Western Black Sea (2) and Northeast Anatolia (2) regions.
Table 13. Efficiency Status of FHCs in Group D according to NUTS-1 Level
Level 1 Regions
İstanbul
West Marmara
East Marmara
Aegean
Mediterranean
West Anatolian
Central Anatolian
West Blacksea
East Blacksea
North East A.
Middle East A.
South East A.
Efficient
-
Mean
Inefficient Efficien
cy(%-
54,40
67,23
66,47
64,86
69,55
64,50
60,41
57,52
55,10
61,87
59,19
75,91
Number
of FM-
Numb
er of
Family
Health
Perso Number Number
nnel of FMU of AHK-
-
-
The regions with the highest mean efficiency score in Group D are the Southeastern
Anatolia (75.91%), Mediterranean (69.55%) and Western Marmara (67.23%) regions
and the regions with the lowest mean efficiency are the Istanbul (54.40%), eastern
Black Sea (55.10%) and Western Black Sea (57.52%) regions. In order to be able to
operate the units that are inefficient in this group efficiently, the number of AHK and
FMU inputs should be focused and improvements should be made. Improvements in
FM and Family health personnel numbers in all regions in this group have a very weak
impact on the efficiency levels of the units. NUTS-1 level Thematic Map created
according to the mean efficiency values of FHCs in Group D is presented in Map 4.
Map 4. Thematic mapping of FHCs in Group D according to NUTS-1 Level
Group E
According to the NUTS-1 level of the FHCs in Group E, the intervention areas that can
provide efficiency and technical efficiency status are given in Table 14. In Group E, the
regions with the most efficient units were Mediterranean (13), Southeastern Anatolia
(13) and Eastern Marmara (10), and the regions with the least efficient units were
Istanbul (1) and Eastern Black Sea (1). The regions where the mean efficiency score
was highest were the Southeastern Anatolia (60,38%), Eastern Marmara (59,41%) and
Mediterranean (57,26%) regions and the regions where the lowest mean efficiency was
the Western Black Sea (42,38%) and Eastern Black Sea (44,29%) regions.
Table 14. Efficiency Status According to NUTS-1 Level of FHCs in Group E
Level 1 Regions
İstanbul
West Marmara
East Marmara
Aegean
Mediterranean
West Anatolian
Central Anatolian
West Blacksea
East Blacksea
North East A.
Middle East A.
South East A.
Numb
er of
Mean
Family
Efficient Inefficient Efficienc
Numb Health
Numbe
y (%)
er of Person Number
r of
FM
nel
of FMU
AHK-
-
49,27
54,15
59,41
52,82
57,26
50,45
50,93
42,38
44,29
52,34
50,70
60,38
-
-
-
-
The number of AHK and FMU inputs in units that are inefficient in Group E can be
improved, and the efficiency scores of these units can be increased. The number of FM
and Family health personnel in FHCs in Group E has little impact on improving
efficiency scores compared to others. In map 5, NUTS-1 level Thematic Map was
created based on the mean efficiency values of FHCs in Group E.
Map 5. Thematic Mapping of FHCs in Group E According to NUTS-1 Level
DISCUSSION
Due to the relative measurement of the DEA and the various efficiency scores
according to the variables and models used, the comparisons between the studies
should be very careful. Such a comparison can only be meaningful among the studies
in which the same DMU is evaluated by using the same variables and models and their
efficiency in different years. However, it is more likely to encounter several studies in
practice. Therefore, the discussion part of this study included an evaluation of the
findings of this study and the basic judgments of comparable studies.
In order to be effective, units that are inefficient in all groups should first improve the
number of AHK. One of the most grievous issues of Family Physicians is the fact that
they have the workload they face because of the number of registered persons (World
Bank, 2013). Due to the professional responsibilities imposed on Family Physicians,
family physicians have difficulty in their personal lives (WONCA, 2011). In addition, it is
envisaged that a regulation will be made by the public authority in the number of AHK
by 2023, with 2,000 people falling to each FM (Akdag, 2011, p.370). In addition to the
above, given the patterns of the countries in which the FM system is applied, the fact
that more people are connected to family physicians in Turkey than in other countries
shows that this finding is in place.
Secondarily, the input that needs to focus on inefficient units is the number of FMU,
except for Group C. When the data were analyzed, it was observed that there were at
least 1, at most 19 FMUs are FHC-associated in nationwide. However, in the official
documents, there is a provision that at least two or at most six FMUs should be
connected to each FHC (Official Journal, 25 May 2010, m.19). While arranging FMU
numbers, it is considered that it is possible to connect some units by considering
criteria such as the number of AHK, demographic characteristics of persons, and habits
of use of health services.
The number of family health personnel in all groups, both efficient and inefficient, was
found to be the least beneficial variable in performance. In a study conducted in the
USA, it was found that the number of full-time nurse and physician assistants had an
adverse effect on the technical efficiency scores. (Amico, 2012, p.97). It can be said
that the finding as mentioned above is parallel to the finding in our study. Also, the
particularized performance criteria are considered criteria for the measurement of
Physicians ' performance. For this reason, the number of family health personnel is
thought to be another factor that has limited impact on the efficiency of the units. FM
services to be done through the control and order of the Physician and family health
personnel to transfer the proceedings on behalf of family physicians to Information
System maintains this idea.
Another significant finding in this study is that the efficiency scores differ between
regions. This finding is similar to the findings of a study conducted by Üner (2006) on
117 primary health centers in Denizli. In the study carried out by Üner, it was found that
there was a significant difference between the region and the technical efficiency
scores of the primary health centers (Üner, 2006, p.114). In a study by Erinç (2013)
evaluating the efficiency of Family Physicians in Sinop province, it was found that the
efficiency performances of family physicians did not differ according to a region. In our
study, it is very reasonable to find differences between regions since all FCHs in
Turkey are taken into consideration.
In this study, it was observed that the environmental characteristics of FHCs have an
essential impact on the efficiency. When taking into consideration the efficiencies of
the regions, while the units serving in the Southeastern and Eastern Anatolia regions
were more efficient, the efficiency levels of the units serving in the Central Anatolia
region were lower. This situation is thought to be due to the population characteristics
of the given region. According to the data of the Turkish Statistical Institute, the
provinces with the highest birth rates are placed in the Southeastern and Eastern
Anatolia regions. Due to the determination of legal performance criteria for pregnancy
and Child Health Services, efficiency scores of the Southeastern and Eastern
provinces are higher. In the study conducted by Cordero-Ferrera and others (2014), it
supports this idea to state that the increase in population density (usually in big cities)
has led to more efficient units. In the provinces to the North and East of Central
Anatolia region, it is thought that the birth rates are close to the lowest levels and that
the elderly population in these regions is higher especially in the rural areas and that
the efficiency levels of FHCs in these regions are lower in our study.
CONCLUSION
In this study, the technical efficiencies of FHCs, which is a newer unit in the Turkish
health system compared to other units, were evaluated and it was aimed to define the
potential improvement fields that should be focused on by public administrators in order
for the units to work efficiently. In this context, a total of 7.145 FHC, which operates in
2015 within the borders of Turkey, was separated into five groups determined by the
Ministry of Health and analyzed with the help of the DEA. According to the results of
the analysis, approximately 3%-5% of the FHC in each group were efficient, and the
mean efficiency values of the groups ranged between 53.00% and 71.80% and the
efficiency scores differ statistically from the groups. It was found that the primary
variable that causes FHCs to be efficient or inefficient is the number of AHK variables
and the secondary significance variables are varied according to groups, and the
number of FM and FM variables are determined. The number of family health
personnel was found to be the least effective variable in the efficient or inefficient
functioning of the units.
The first obligation to be done by the public administration is to rearrange the AHK
numbers. For this reason, the employment of the new Family physicians, the other
units of the family physicians more than the absent units to split-shift, and the orders of
the assigned transactions made by the family physician by employing family health
personnel within the framework of the management of authority are the main
suggestions. The limited number of trained family physicians makes the employment of
new family physicians difficult. Instead, the redistribution of Family Physicians and the
employment and empowerment of family health personnel have a more likely
application field in policy applications. In this way, family health workers will be able to
work more efficiently in the system. However, it should not be neglected that a small
adjustment to be made at this point can have a significant impact across the country.
This study has some limitations: First, the grouping of decision-making units was based
on the grouping determined by the Ministry of health, not on the analytical method.
Second, the findings and results obtained within the scope of the study should be
evaluated by taking into account the decided variables and models. Otherwise, false
comparisons can be made. Finally, although it is planned to take the data between
2010 and 2015 within the scope of the study, it is stated that the most up-to-date and
most credible data available across the country is in 2015 as the FM system is a newly
established system. Therefore, only 2015 data were shared by the public
administration. Therefore, there was only a one-year performance evaluation.
Suggestions for Future Research
Since the services and performance criteria offered may change over the years, the
current performance criteria can be taken into consideration in future studies and
evaluated together for various years. Thus, total factor productivity (such as the
Malmquist Productivity index) can be calculated in which changes over the years can
be analyzed.
Since the FM system is a subunit of Public Health Services, an efficiency index can be
developed in which the units such as FM units connected to Public Health Services,
Public Health Center, Public Health Laboratories can be weighted and evaluated
together from a more holistic perspective.
When the studies in primary health care are evaluated in two categories with
methodology and application focus, it is clear that this study is an application-oriented
study. Although studies of primary health care have been conducted consistently with
the DEA, it can be asserted that the methodology in this field is in development and
therefore needs to be improved to meet the complex production characteristics of
primary health care. The evaluation of other performance aspects such as quality,
productivity, and innovation in the new method will help administrators to make more
accurate decisions on the issues related to efficiency.