Linking ‘Soft’ With ‘Hard’: A Case of Health System
Planning in a Developing Country
Shams-ur Rahman
Graduate School of Management
University of Western Australia, Nedlands, WA-6907, Australia
Tel: 61408-9380 1451, Fax: 61408-9380 1072
srahman @ecel.uwa.edu.au
and
David K Smith
School of Mathematical Sciences
University of Exeter, Exeter, EX4 4QE, UK
Tel: 444+392-264478, Fax: 444+392-264460
d.k.smith @ex.ac.uk
Abstract
In recent years, Bangladesh, like many developing countries, has given high priority
to the development of primary health care. The twin aims of the programme are to
extend coverage of health services in rural areas and improve the overall health of
the country's population. In 1993, the European Commission (EC) initiated a project
called the Thana Functional Improvement Pilot Project (TFIPP) which is providing
assistance to the Ministry of Health and Family Welfare (MOHFW) of the
Government of Bangladesh. The objective of the TFIPP is to develop an efficient and
effective operational system to deploy a health service in the country designed by the
Government and the World Bank in partnership. The TFIPP consists of four
functional units and one of these is the Operations Research and Social Development
(ORSD) unit. The ORSD is the youngest unit among all and was formally created in
1997. Since then it became an integral part of the TFIPP. This study describes how
the ORSD unit applied OR models in conjunction with systems methodology to locate
health facilities at the community level. First systems methodology was applied to
develop knowledge and understanding about the problem, to identify variables
interacting in the health system and to determine what affects the quality of services.
Some of these variables were then considered to formulate location models as a
maximum covering location problem (MCLP) so as to find suitable sites for health
facilities in rural Bangladesh.
Background
In Bangladesh, a unit of local government is known as a “thana”. Typically, a thana
has a population measured in tens to hundreds of thousands. Each thana consists of
about ten to fifteen “Unions”. In 1993, the European Commission (EC) initiated the
Thana Functional Improvement Pilot Project (TFIPP). The EC-sponsored project is
providing assistance to the Ministry of Health and Family Welfare (MOHFW) of the
Government of Bangladesh. Recently, the Government, in conjunction with the World
Bank, has developed a comprehensive health policy to extend the coverage of primary
health care. The objective of the TFIPP is to develop an efficient and effective
operational system so that the partnership of the Government and the World Bank
can deploy a health service in the country. As its name implies, the TFIPP operates at
the Thana level and below. The relationship between the Government, the World
Bank, the EC and TFIPP is shown in Figure 1. The TFIPP consists of four functional
units, which are working in an integrated manner in order to achieve its objective.
The functional units are:
Training and Communication
Public Health
Management Intervention, and
Operations Research and Social Development (ORSD) unit.
ah Gort
Development
and Extension of
Health Services
J
European 1} Government of |———>}
Commission (EC) |¢. Bangladesh le
| |
TFIPP (Pilot MOHFW
Project)
World Bank
FIG. 1: Health policy development and implementation bodies
The ORSD is the youngest unit among all and was formally created in 1997. Since
then it has become an integral part of the TFIPP. The unit is making plans to ensure
that there is an equitable and efficient system for the delivery of health facilities
throughout the country. For some considerable time, the Government’s national
health policy has been based on a health care system with three levels of services
from primary health care to specialised care. Primary health care includes treatment
of common diseases, preventive and promotive health care at community/village,
union, and thana levels. The health delivery system at thana level and below consists
of a Thana Health Complex (THC), Health and Family Welfare Centers (HFWCs)
and Community Clinics (CCs). Figure 2 shows the administrative divisions at the
national level and health facilities at each level. This study is concerned with the
location of CCs at the community/village level. The functions of this type of facility
is to deliver a mix of health services which is called the Essential Services Package
(ESP). This includes provision for reproductive health, child health, communicable
disease control, limited curative care, and behaviour change communication.
Tertiary
Care Nation
‘& Region
SHMCH
District
District Hospitals
Secondary
Care
Thana THe
HFWC
Union
Primary
Care
Community/Village Comamnity, Cale
‘Type of Care Administrative Levels Health Facility
Note: SHMCH - Speciatized Hospital and Medical College Hospital
THC - Thana Health Complex
HFWC - Health and Family Welfare Centre
FIG.2: Hierarchy of public health care system in Bangladesh
This study describes how the ORSD unit applied location-allocation models in
conjunction with systems methodology to locate health facilities at the community
level. The decision making process involved in the study is shown in Figure 3. First a
systems approach was applied to develop conceptual maps of the health services
planning problem. The purpose was to develop knowledge and understanding about
the problem, identify variables interacting in the system and determine what affects
the quality of services. Some of these variables were then considered to formulate
location models so as to find suitable sites for community clinics.
— NOON
Issue: health
Locational
modelling
Conceptual
services planning maps
NE NN”
FIG 3: Blending ‘soft’ and ‘hard’ approaches within the health services planning problem.
The Conceptual Maps of the Health Services Planning Problem
Between February 1995 and October 1996 TFIPP conducted a community survey
amongst about 12,000 people. The purpose of the study was to collect baseline data
and information on health service knowledge, attitude and practice. The ORSD has
also conducted a number of studies specifically aimed at understanding peoples’
attitude towards the public health services (Batlama, 1997), to investigate the
relationship between poverty level, health service knowledge and utilization of health
services (Anwar, 1997, Saha, 1997), and to determine the effect of distance on the
utilization of ambulances (Anwar et al., 1998). Based on these items of research, the
ORSD unit has developed a conceptual map of demand and supply of health services
in rural Bangladesh. There are four key elements in the conceptual map (Figure 4)
which are:
Management of health
services resources
Education and knowledge
about health and health
services
¥
SUPPLY
z
DEMAND
Affordability
(Cost/time)
Distribution/availability
of static health facilities
FIG. 4: The conceptual map of the health services planning problem
« Education and knowledge of rural people about health and health services.
« Management of resources related to health services.
¢ Distribution/availability of static health facilities.
¢ Affordability (cost/time).
This map has been further elaborated in Figure 5. The purpose of the mapping process
was to understand the problem and develop knowledge about the interaction effect of
the variables in the system. The elements such as education and knowledge,
distribution of health facilities and affordability jointly determine the level of demand
for health services. The management of the health resources element directly and
indirectly through the distribution of health facility element determines the level and
quality of supply of health services. While these elements affect one part of the
system more than the other, the element of the distribution of health facilities affects
both supply and demand. Through the mapping process the indirect impact of
education and the direct effect of the management of health resources on health
services became apparent to the decision makers. Most of the variables of these
elements cannot be quantified and therefore intervention programmes and accountable
management systems would improve the quality of health services. Some variables
can be rationally quantified and these variables can be used for location modelling
taking into account the government health policies. In this study an attempt has been
made to model the health facility deployment system at the community level
incorporating variables such as travel-distance, affordability, location of existing
facilities and availability of other infrastructural facilities and services. Therefore we
concentrate on the community clinics.
Dabir Ali (The Decision Maker)
Ba
Understanding
Availability
of Doctors!
Paramedics,
of Resources
SUPPLY
FIG 5: The extended conceptual map of the health services planning problem
Location-allocation models
Where should the community clinics be located? In Bangladesh, as in many
developing countries, there is a strong negative correlation between the distance from
a place where health provision is available and the rate of utilization of the service
(Khan, 1988, Rahman ef al., 1982). Accordingly, the location problem is more
appropriately considered as a maximum covering location problem (MCLP) (Church
and ReVelle, 1974) to try and ensure that as many people as possible are within a
given distance of facilities. The objective ina MCLP is to maximise the population
who are expected to use the facilities, and this is measured by the number who live
within a given radius around a facility which provides health care using a fixed
number of facilities. To try and locate the best sites, we have used the Teitz and Bart
(1968) heuristic for facility location.
Government Policy and Location of Community Clinics
In a recent document, the Government of Bangladesh (MOHFW, 1998), in
conjunction with the World Bank, developed some policy guidelines as to how rural
health facilities be deployed. These guidelines relate to physical accessibility and
population coverage.
Physical Accessibility: The Community Clinics (CCs) must be accessible within half
an hour travel-time. It is perhaps simpler, from the point of view of modelling, to
think in terms of travel-distance rather than travel-time. In rural Bangladesh most
people walk to service facilities. So, for most people, there is a direct correlation
between time and distance. Therefore, the maximum travel-time was translated by
TFIPP into a maximum travel distance. Many studies on location of facilities in
developing countries have used these two, (almost) equivalent, metrics. Distance can
be easily measured, and studies of the behaviour of patients allows this to be
translated into a travel time. It must be remembered that in an agricultural
community, attending a health centre means both the loss of income and the loss of
time from one's land, which affects the whole family, particularly at certain times of
the agricultural calendar. In the case of CCs, the maximum travel distance (S) was
considered to be 2 km. However, in the analysis more than one such upper limit on
travel-distance was considered in order to find its effect on the population coverage
and number of facilities.
Population Coverage: With respect to the level of coverage by one CC, the
Government has suggested three service delivery options. These are:
1. One CC per 1500 population
2. One CC per 3000 population, and
3. One CC per 6000 population.
The coverage issue was discussed amongst Government planners and decision-makers
in a consensus building workshop, where the task was to identify the best service
delivery option. However, no concrete suggestion came out as to which alternative
option is to be considered for implementation. It would be appropriate to argue that
the coverage issue should not be looked at in isolation, rather be analyzed in
conjunction with the maximum travel-distance.
In the analysis which follows, in addition to the Government’s policy guidelines, we
also employed the following assumptions and constraints:
Village size: While carrying out the survey in the study area it was evident that neither
the local people (users) nor the health authority (providers) would accept location of
facilities in very small villages. Locating desirable facilities where a village
population exceeds some threshold has often been used in similar studies of
developing countries (Patel, 1979, Moore and ReVelle, 1982). There is no set rule or
guideline of the Government of Bangladesh on the population that a village must have
to be chosen as a facility site. However, the previous locational decisions show that
all the existing CC sites had more than 1000 people. In this study we have considered
a village as a feasible facility site for a CC provided it has a minimum population of
1000 people.
Uncapacitated facility: CCs are meant to organize immunization activities, treat
diarrhoeal diseases and fever cases, and work for the family planning programmes in
the rural areas. Since these are outpatient facilities, no constraint has been imposed
on their capacity. However, in this study, the variation in the demand for services
among the facilities has been reduced by including a minimum population constraint
for a village to be a potential centre. Since it is uncommon for several large villages
to be close together, the demand will tend to come from one large village and the
sparser population around it. Any further variation of workload could be handled by
allocating health personnel according to the demand at the facilities.
Other facilities: In addition, the presence of electricity and acceptable water supplies
were considered as other criteria for feasible CC sites.
Study Area Profile
The initial plan of this study was to deal with the problem of locating Community
Clinics in three unions such as Lokhpur union (Thana Fakirhat, District Bagherhat),
Pairaband union (Thana Mithapukur, District Rangpur) and Bakshimul union (Thana
Burichaung, District Comilla). However, because of data problems, the locational
analysis of CCs was conducted only at Lokhpur union. It is one of nine unions in
Fakirhat Thana, situated towards the centre-western part of the Thana and only 8 kms
from the division headquarter Khulna. Its close proximity to metropolitan Khulna and
a good communication with Bagerhat and Port Mongla has given the union a different
characteristic with respect to socio-economic acitivities. The study area has a total
population of about 17 thousand persons of whom about 36% are considered to be
literate. Agriculture is the main economic activity in this area. At present, one HFWC
and six CCs provide health care to the local people. There are 9 primary schools and 2
high schools in the study area.
Lokhpur union consists of nine mouzas/villages. The mouzas were identified on a 1:8-
scaled map produced by the Health Assistance of Lokhpur union. The networks of
roads and rural footpaths in the area was updated on the map with the help of the head
surveyor. Each link on the network was identified and measured and this data was
used to compute a 9x9 distance matrix using a shortest path algorithm.
Locational Analysis of Community Clinics
Taking into account the government policies and other constraints (mentioned above),
the following analysis was done:
1. Maximum travel distance = 2 kms and minimum population = 1000 people.
Population Coverage
120
100
% Covered
yy
1 2 3 4 5
No of facility
FIG. 6: Trade-off curve
The results of the analysis is shown in Figure 6. It shows that only 4 CCs are required
to cover 90% of the population in the union within 2 kms travel-distance. An extra
CC will supply health care to the remaining population. With these solutions, the
workload at the clinics varies between approximately 2400 and 4800 people.
2. Maximum travel distance = 3 kms and minimum population = 1000 people.
The results are presented in Figure 7. The health delivery system with only 2 CCs
would be able to cover about 96% of the population.
Population Coverage
120
100
0 ee.
% covered
=)
1 2 3
No. of Facility
FIG. 7: Trade-off curve
Maximum travel distance = 1.6 kms (1 mile) and minimum population = 1000
people.
The results are presented in Figure 8. The optimal location of 6 CCs can cover about
96% of the population in the union.
Population Coverage
% Covered
8
ol
~
1 2 3 4 5 6
No. of Facility
FIG. 8: Trade off curve
The analysis shows that if travel-distance is increased by 33% (from 2 kms to 3 kms),
the number of CCs is decreased from 5 to 2 for approximately the same population
coverage. Also, when the maximum travel-distance is reduced by half (from 3 kms to
1.6 kms), the number of CCs required increases from 2 to 6 for the same population
coverage.
Conclusion
No method or methodology is able to offer a complete view of the complexities faced
in projects. Each may provide insights that are useful for reflection and action. Many
researchers have demonstrated that the blending of ‘soft’ and ‘hard’ methods produce
richer picture for understanding complex relationships which leads to a better
decisions (Pidd, 1996, Ackermann et al., 1997, Wolstenholme, 1999). In this study
OR models were applied in conjunction with systems methodology to locate health
facilities at the community level. First a conceptual map as a systems methodology
was developed to learn about the problem, to identify variables interacting in the
health system and to determine what affects the quality of services. Some of these
variables were then considered so as to formulate a maximum covering location
problem (MCLP) as an optimisation model to find suitable sites for health facilities in
rural Bangladesh.
The analysis has suggested several sets of solutions with respect to the location of
CCs for efficient delivery of heath care. Clearly, the objective of the study was not to
find a single optimal decision, rather to develop and test feasible decision processes in
the light of the government’s health policies. The results, therefore, should be
regarded as an aid to the local health planner’s intuition and not as a total replacement
for it.
References
Ackermann, F. Eden, C. and Williams, T.M. (1997). Modeling for litigation: mixing
qualitative and quantitative approaches. Interfaces 27, 48-65.
Anwar, I. (1997). Socio-demographic factors in utilization of government health
services and policy recommendations. Unpublished Report, TFIPP, Dhaka,
Bangladesh (contact first author for a copy).
Anwar, I., Batlama, J. and Groenendijk, K. (1998). Review of ambulance operations
in TFIPP Thanas. Unpublished Report, TFIPP, Dhaka, Bangladesh (contact first
author for a copy).
Batlama, J. (1997). Attitude towards health services: how much difference does it
make? Unpublished Report, TFIPP, Dhaka, Bangladesh (contact first author for a
copy).
Church, R.L. and ReVelle, C.S. (1974). The maximal covering location problem.
Papers, Regional Science Association 32, 101-118.
Khan, M.R. (Ed.) (1988). Evaluation of Primary Health Facilities and their
Limitations Specially in the Rural Areas of Bangladesh. Research Monograph No 7,
Bangladesh Institute of Development Studies, Dhaka, Bangladesh.
MOHFW (Ministry of Health and Family Welfare, Government of Bangladesh)
(1998), Fifth Health and Population Programme 1998-2003 (HAPP-5) — Programme
Implementation Plan, January, Dhaka, Bangladesh.
Moore, G.C. and ReVelle, C.S. (1982). The hierarchical service location problem.
Management Science 28, 775-780.
Patel, N. (1979). Locating rural social service centres in India. Management Science
25, 22-30.
Pidd, M. (1996). Tools for Thinking: Modelling in Management Science, John Wiley
& Sons, England.
Rahman, M.M, Aziz, K.M.S., Munshi, H.M., Patwari, Y. and Rahman, M.A. (1982).
Diarrhoea clinic in rural Bangladesh: influence of distance, age and sex on attendance
and diarrhoeal mortality. American Journal of Public Health 72, 1124-28.
Saha, R. (1997). Poverty and health service utilization. Unpublished Report, TFIPP,
Dhaka, Bangladesh (contact first author for a copy).
Teitz, M.B. and Bart, P. (1968). Heuristic methods for estimating the vertex median
of a weighted graph. Operations Research 16, 955-961.
Wolstenholme, E.F. (1999). Qualitative and quantitative modelling: the evolving
balance. Journal of Operational Research Society 50, 422-428.
CC BY-NC-SA 4.0