1
2
3
4
5
Preventive Care Resource Allocation in Developing Countries: Can Rational
Planning Techniques help in Allocating Vaccinators in Dera Ismael Khan
(DIK) District of Pakistan?
6
Preventive care service delivery faces greater challenge when it comes to service effectiveness.
7
Within the context of preventive care, child immunization in developing countries generally
8
lacks strategic planning.
9
distributed hardly meeting the need. This paper presents a case of applying strategic
10
management tools for child immunization in Pakistan. Bearing in mind effective and equitable
11
delivery of child immunization service delivery we explore application of integer programming
12
technique to support Expanded Programme of Immunization (EPI) service in D.I.Khan District
13
of Pakistan. The work here concentrates on decentralization of resources through rational
14
planning tools in order to localize preventive care service delivery that is easy, cost effective and
15
equitable. Two aspects of service delivery are given importance, 1) covering the target
16
population to the highest possible level of vaccination and 2) ensure equality among population
17
scattered over geographical areas especially in rural dwellings. In this regard resource allocation
18
mechanism assumes two alternative ways of delivering the EPI service i.e. a) health district is
19
subdivided in to localities and vaccinators may be allocated to vaccination centres in various
20
localities to visit and vaccinate children within their administrative boundaries, and b) within
21
the localized planning system, vaccinators may administratively be allocated to health facilities
22
in various localities but operationally they may visit and vaccinate children across their
23
administrative boundaries subject to saving in travel time. In both cases a trade off between
24
travel timesaving and equality in service provision has been examined.
Primary health care facilities in developing countries are thinly
1
25
Key Words:
26
decentralization, localized planning, integer programming, travelling salesman.
27
1.1)
Preventive care, resource allocation, vaccination service, primary health care,
Introduction
28
There is no doubt that national health predominantly dependents upon preventive care.
29
This implies that within the domain of primary and secondary health care more emphasis
30
should be on the delivery of preventive care in order to lessen the burden on secondary health
31
care. Health care need is multidimensional therefore social planners have difficult task on
32
hand to distribute resources within the social sector that help integrating social sector services
33
and within health services subdivide resources appropriately between primary and secondary
34
care. One solution we see in the literature is decentralization of primary health care services
35
so that various services complement each other and benefits reach the grassroots effectively.
36
According to one study in Manitoba, Canada, components of preventive care delivery were
37
examined by three different methods; a) childhood immunizations (by physicians and public
38
health nurses under a government program), screening mammography (through a government
39
program introduced in 1995), and cervical cancer screening (no program) [1]. In case of
40
Manitoba the purpose of study was to understand the effect of socioeconomic status on the
41
use of various components of preventive care which is different than what we are focusing in
42
this research. Uniqueness of our approach will become further clearer as we will, later in this
43
paper, highlight evidence from the literature where cost of immunization has been the
44
elementary issue not the delivery tactics. The study of impact of socioeconomic status on the
45
use of service or cost behaviour of vaccination do have the underlying objective of identifying
46
needy groups and delivering immunization service equitably but tactical planning for service
47
delivery has not been a major concern in the literature which is a core issue in this paper.
2
48
1.2)
Objectives
49
Main objective of this research is to demonstrate the application of operations
50
management techniques in a real life scenario of a developing country for preventive care
51
planning. In this respect our paper focuses on the delivery of immunization service through
52
decentralized primary health care in Dera Ismail Khan (DIK) - a health district of Pakistan. The
53
scope for decentralized primary care is not a main issue of this paper however it is evident that
54
decentralization with reference to Pakistan is not a new concept [2]. Like primary care in
55
Manitoba, the primary care in DIK District also runs several programmes in parallel e.g. Lady
56
Health Workers programme to provide support for child and mother, National Programme for
57
Family Planning, Expanded Programme for Immunization (EPI), Malaria Control
58
Programme, National AIDS Prevention and Control Programme, and Tuberculosis Control
59
Programme. Each of these programmes is planned by a mixture of door to door visits and
60
clinical surgeries to deliver the service.
61
As mentioned earlier with reference to providing vaccination service the major
62
concern in the literature has been the cost of providing service. For example a study
63
concentrating on the variation in the cost of delivering routine immunization service in Peru
64
recommended generalizing their findings of cost variation for use in other settings to support
65
decision-makers in their attempt to plan [3]. The main focus was providing disaggregated
66
data on local costs which planners normally do not have [3]. Our stance however is the
67
tactical planning for the delivery of service with equity that we believe would take care of
68
cost distribution among localities.
69
In line with this the current paper shows application of advanced planning methods to
70
allocate vaccinators for EPI. We have chosen this service; firstly because vaccination service in
3
71
DIK is provided through door to door visits and it would be possible to apply travelling
72
salesman models easily and secondly preventing infants from infectious diseases should be
73
prioritized to give them a good quality life. Application of integer programming models has
74
been demonstrated to allocate district nurses to localities in the UK [4, 5] therefore in the
75
current case our focus is to show how integer programming can be used for an analogous
76
situation in a developing country.
77
According to general perception infectious diseases are a major problem in the developing
78
countries. Thousands of children die of Measles, Pertussis, Poliomyelitis, Tuberculosis, Diphtheria, and
79
Tetanus.
80
complications of these maladies. Therefore vaccination priority over all other preventive measures to
81
give children a chance of good quality living. So is evident from the practices in Pakistan that to
82
vaccinate children under five against above six diseases the district health authority deploys vaccinators
83
to visit children in their houses to immunize them under EPI (Expanded Programme of Immunization)
84
scheme.
85
2.1)
Many more are crippled, blinded and spend the rest of their lives with one or more
Why Integer Programming?
86
Before going to the integer models for EPI resource allocation it is worth mentioning
87
that Integer Programming Techniques are widely used in large-scale strategic planning
88
decisions where circumstances require planning models to contain integer-valued variables.
89
For example, in utilisation of an aircraft, a ferry, or any other piece of equipment that
90
provides large-scale capacity and is expensive-a fractional value is normally meaningless in
91
the context of any decision problem. Similarly planning problems where fixed costs/set-up
92
costs, batch sizes, or either-or decisions are involved can be solved using integer
93
programming models. Many other decision problems involving combinatorial optimization
94
such as travelling salesman, machine scheduling, sequencing etc. with resource constraints
4
95
can be solved using integer modelling [6, 7, and 8]. Analogous to that of the travelling
96
salesman approach, the present paper offers two mixed integer programming models to
97
allocate vaccinators for primary health care service delivery in DIK District.
98
2.2)
Input Data
99
In order to apply integer programming model input data required are the need for
100
vaccination and estimated travel time between various need and supply points. Data
101
collection is further explained as follows.
102
2.2.1) Need for Immunization Service
103
The method of calculating need for the vaccination service is much simpler than that
104
used for any other primary health care service. It is fairly straight forward because each child
105
that is born has to be vaccinated. Therefore the 'criterion of need' calculation is based on the
106
population of children. However it is necessary to know which age categories are to be
107
considered and how immunization is to be carried out during childhood.
108
The most common schedule of immunization deals with two age categories of
109
children; one is 0-1 year and second is 4-5 years of age. In a standard practice between 0-1
110
year of age immunization involves five stages of vaccination which means that each child in
111
this age category has to be visited five times in a year. This implies that the need for
112
immunization in this age category is five times the population of this age category. Similarly
113
the need of children between 4-5 years however is simply the same as the total population in
114
that age category.
115
Based on this criterion the age specific need for immunization has been worked out on
116
union council bases of the three localities of DIK [2] and used as input for our models.
117
Administratively DIK is suggested to have three localities for health planning i.e. Paroa, Dera,
5
118
and Paharpur further subdivided into union councils, the smallest unit of a district. The need is
119
identified on union council basis. The need data is however not given here for space saving.
120
2.2.2) Travel Time Estimation
121
The second data set required to allocate vaccinators is the estimated travel times
122
between vaccination centres and union councils. In DIK District there are reportedly sixteen
123
vaccination centres which deliver the immunization service throughout 25 union councils
124
(including municipal committee area as well). The estimated travel times between vaccination
125
centres and union councils were calculated using the DIK road network map. All union councils
126
and the municipal committee areas are interconnected by roads, most of which are metalled.
127
However some union councils are not directly linked by metalled roads, in such cases
128
unmetalled roads or ordinary tracks are used. In order to estimate travel times we have
129
measured distances on metalled roads and on unmetalled and ordinary tracks, where
130
appropriate.
131
Within each union council, most of the population resides within the vicinity of the
132
union council's headquarters although there are number of small villages scattered nearby. We
133
have therefore estimated the travelling time between union council headquarters and vaccination
134
centres. The method used is as follows;
135
1) All the union councils and vaccination centres were located on the road network map.
136
2) With the help of a map measure, distances between union councils and vaccination
137
centres were measured. For each vaccination centre and union council, distances on
138
metalled and where appropriate on unmetalled and ordinary track were taken
139
separately in kilometers.
6
140
3) Based on these distances motor bicycle travelling times were calculated. Motor
141
bicycle travelling time was taken because vaccinators are mostly provided with motor
142
bicycles for this purpose (although within the city vaccinators may walk or use
143
bicycles).
144
4) Assuming an average speed of 30 km/hour on metalled roads and 10 km/hour on
145
unmetalled or ordinary tracks, travel times both on the two road surface types were
146
obtained separately in minutes.
147
5) Finally the travel times on metalled roads and unmetalled or ordinary tracks were
148
combined to give total travel time between each vaccination centre and each union
149
council in minutes. A matrix of travel times between vaccination centres and union
150
councils was developed and used as input data for both the models. This matrix has
151
not been given here for space saving.
152
The data on need and travel times was then used to carry out an exercise to allocate vaccinators
153
to localities. This is explained as follows.
154
2.2.3) Allocation of Vaccinators
155
In DIK as mentioned earlier there are sixteen vaccination centres and 25 union councils.
156
Vaccinators start their day's work from the vaccination centre by collecting vaccines. To keep
157
vaccines at the required temperature vaccinators are provided with a special ice boxes. As
158
mentioned earlier in this Paper, the delivery of vaccination service in DIK is analogous to the
159
district nursing service in Tower Hamlets; therefore the integer programming models already
160
explained in [4, 5] have been adapted to the problem of the allocation of vaccinators.
161
We have assumed that the decentralized service may be delivered in two ways i.e. 1)
162
strictly locality bound service, and 2) flexible service allowing service across the board. This has
7
163
allowed us using two models. In Model one we have assumed that there are three localities in
164
DIK each comprising a set of vaccination centres and union councils. It is assumed that the
165
service delivery is planned to take place strictly within the prescribed locality boundaries -
166
disallowing cross boundary flows. Model two considers that localities serve primarily as
167
managerial units for the set of vaccination centres located within them however the vaccination
168
centres can serve any set of union councils throughout the tehsil - allowing cross boundary
169
flows. The objective of analyzing the situation in two ways is to examine the implications of
170
removing locality based restrictions on equitable service delivery.
171
The following assumptions are made:
172
a) On an average a vaccinator can conveniently visit and vaccinate 5 children per day.
173
The number of children assumed to be vaccinated per day is unexpectedly low
174
because vaccinators have to visit children in their homes and because of travelling
175
difficulties and improper roads within each union council. With this assumption we
176
will see that both models suggest that, with the current level of provision, the need
177
that can be satisfied will remain below 59%. Looking at the current situation of health
178
in Pakistan this result seems fairly reasonable and would provide a base for EPI to
179
cover more population.
180
181
182
183
184
b) There are 273 working days in a year (365 less 52 weekly offs, less 40 public holidays
and other leaves); and
c) Travel time between vaccination centres and union councils has been taken as a
measure of cost.
The explanation of the two models is given below.
8
185
3.1)
Model One
186
Model One focuses on service deliver strictly within the context of existing localities,
187
aims to allocate an integer number of vaccinators to each locality so that the proportion of
188
need met is as equal as possible across all union councils and across all children to be
189
vaccinated (the equity constraint), and so that the total travelling time of vaccinators is
190
minimized. Full description of the model is given in Appendix 1.
191
3.2)
Model Two
192
In this model, it is assumed that localities remain as managerial units for the vaccination centres within
193
their boundaries, but vaccinators from any vaccination centre are able to be allocated to serve the populations of
194
any union council within the district. This assumption has been made to examine the implications of removing
195
locality based restrictions on service delivery. The aim is the same as in Model One, to allocate an integer
196
number of vaccinators to each locality in such a way that the proportion of the need that is met is the same (the
197
equity constraint), and that the total time travelled by vaccinators is minimized. Full description of the model two
198
is given in Appendix 2.
199
Both models are similar in that their objective is to allocate vaccinators to localities.
200
However they solve the problem by assuming two different methods of service delivery.
201
Model One disallows cross boundary flows whereas Model Two relaxes this condition. The
202
models were run on the mathematical programming system Sciconic and the solutions
203
obtained are explained below.
204
4.1)
Solution of Model One
205
Model One, in which service is delivered strictly within the context of proposed locality
206
boundaries, gave a feasible solution when 3% deviation from equity was allowed. The model
207
was then solved repeatedly allowing deviations from equity of 5%, 10%, 15%, 20%, and 25% in
9
208
order to examine the impact on the allocation of vaccinators to localities and on travel time. A
209
summary of results is given in the Table 1.
210
Table 1 reveals that with a 3% deviation from equity, the need that could be satisfied for
211
each union council and each age category is within the limits of 58.6% and 55.6%. The table
212
also shows that as the deviation from equity increases, the distance travelled reduces; so
213
permitting equity to vary by 25% rather than 3% results in a 14.5% saving in travel time.
214
Otherwise it is evident that the overall allocation pattern to localities does not show any major
215
change.
216
Although 14.5% is a considerable reduction in travel time, it must be set against the
217
importance of child immunization. The two dimensions are incommensurable, but priority must
218
clearly be to the latter, suggesting that vaccinators should be allocated with a minimum feasible
219
deviation from equity.
220
The Model has the capability to allocate vaccinators to the localities and individual
221
vaccination centres to serve each union council and each age category. Here for the sake of
222
simplicity we have not gone into such details.
223
4.2) Solution of Model Two
224
Model Two considers the allocation of vaccinators in a situation where localities comprise only a
225
set of vaccination centres which can serve any set of union councils throughout the tehsil. Results
226
obtained from the solution of Model Two are given in Table 2.
227
Table 2 shows that the travelling time in Model Two is 14.2% less than that in case of
228
Model One with a 3% equity deviation. There is 57.1% need satisfied equitably in each union
229
council and in each age category. This means that by disregarding locality boundaries and
230
allowing cross boundary flows there may be reasonable saving in travel time.
10
231
As in case of Model One the Model Two also has the capability to allocate vaccinators to
232
localities and vaccination centres to serve each age category in each union council. For the sake
233
of simplicity this data is not presented here however this is worth to mention that given the
234
proposed localities, this solution allows some cross boundary flows. In order to reach a decision
235
between implementing the solution of either Model One or Two, interaction with the
236
responsible decision-making body would be required. It is interesting to observe that the
237
solution obtained by Model Two in particular is quite close to the actual pattern of vaccinator
238
deployment on the basis of proposed localities, as can be seen in Table 3.
239
Integer programming models have explored useful alternatives for the rational allocation
240
of vaccinators for preventive care in DIK. From above results we are of the view that rational
241
planning methods applied interactively can contribute to the delivery of immunization service
242
that is equitable and cost effective. Such approach can be easily generalized and applied in any
243
situation where service is delivered by visiting the recipient of the service as well as allows
244
planners to find trade-off between alternate organizational tactics as we have shown in case of
245
decentralized care.
246
5.1)
Conclusion
247
This Paper reports the work which is first of its kind to demonstrate the impact of
248
applying strategic planning technique on effective and equitable delivery of preventive care
249
(child immunization) in DIK. The allocation pattern resulting from the integer programming
250
models was not out of line with the actual deployment of vaccinators at the time of data
251
collection.
252
developing countries is a major impediment for the application of rational planning tools
253
nevertheless countries are making progress in this direction as we see efforts being made to
Although availability of reliable data from Pakistan as well as from many
11
254
establish Health Management Information System in Pakistan [9, 10]. From the literature it is
255
also evident that there are adequate sources of data available with reference to health, for
256
example Pakistan Medical Research Council (www.pmrc.org.pk), Aga Khan University
257
(www.aku.edu) and many others in public sector, which would enable analysts to develop
258
models with a reasonable confidence.
259
Our work with regard to preventive care planning is only an indicative; a fuller study
260
could be conducted interactively with a responsible decision-making body, perhaps with access
261
to sources of more accurate data. We have also shown here how operational manager can have
262
flexibility in their decision making and find trade-off between equality of service delivery and
263
cost savings provided options are pre set in consultation with local community as well as in
264
coordination with central authority.
265
With increasing emphasis on preventive care in Pakistan and elsewhere [1, 11] there is
266
evidence of gaining momentum by taking cost into consideration whereas our work shows cost
267
effectiveness by shifting towards localized planning for the delivery of service. Although
268
localized planning will evolve over a period of time and there may be several issues to be
269
tackled. Our work here is just a preliminary step in the introduction of rational planning
270
methods which are equally possible to be applied in developing countries as they are used in
271
developed countries.
272
The scope of this work does not end here. By looking at the sub-continent region we can
273
see India’s Annual Report presents latest statistics on health [12]. There is awareness in
274
subcontinent about use of information in decision-making. Expertise is also available in the
275
region and countries can benefit from mutual cooperation. The subcontinent region including
276
India, Bangladesh, Sri Lanka and Nepal all have the same population characteristics with variety
12
277
of talents, technologies and resources available that can be used to apply rational planning
278
techniques in social sector planning and develop their respective nations by enhancing quality of
279
preventive care. There is also even no shortage of intellectual and material input from west.
280
Researchers [13] and organizations like UN and South Pacific Partnership are playing their part
281
to uplift the preventive care system of developing countries by diagnosing problems, sharing
282
information and providing resources. There is needed to take advantage by deploying resources
283
in the right direction.
284
This research opens doors for development of strategic management techniques with
285
reference to social sector planning in health related areas in developing countries and adds
286
knowledge to strategic management techniques to be used as useful tools for planning social
287
sector services in developing countries.
288
References:-
289
1. Gupta, S., Roos, L., Walld, R., Traverse, D., Dahl, M. Delivering Equitable Care:
290
Comparing Preventive Services in Manitoba. American Journal of Public Health.
291
2003; 93: 2086–2092
292
2. Ishfaq, M., Lodhi B.K. Role of GIS in Social Sector Planning: Can Developing
293
Countries Benefit from the Examples of Primary Health Care (PHC) Planning in
294
Britain? Journal of Community Health 2012; 37:372-382, DOI 10.1007/s 10900-
295
011-9454-7
296
3. Walker, D., Mosqueira, N.R., Penny, M.E., Lanata, C.F., Clark, A.D., Sanderson,
297
C.F.B. & Fox-Rushby, J.A. Variation in the costs of delivering routine immunization.
298
Bulletin of World Health Organization 2004; 82(9), 676-682.
13
299
4. Ishfaq, M. Integer programming methods to support decentralized primary health care
300
resource allocation. AAM, University Sains Malaysia and Academy for Global
301
Business Advancement (AGBA), USA, World Congress May 2007, Penang,
302
Malaysia.
303
5. Ishfaq, M. Decentralized Resource Allocation in Primary Health Care: Formal Methods
304
and their Application in Britain and Pakistan. Ph.D. Thesis LSE, University of London
305
1993.
306
307
308
309
310
311
6. Taha, H. A. Integer Programming: Theory, Applications, and Computations.
Academic Press NY 1975.
7. Wagner, H.M. Principles of Operations Research with Application to Managerial
Decisions. 2a edn. Prentice-Hall, Inc.. Englewood Cliffs, N.J,1975
8. Ziont, S. Linear and Integer Programming. Prentice-Hall, Englewood Cliffs, NJ,
1974.
312
9. Ministry of Health. Implementation of Freedom of Information Ordinance 2002: A
313
Case Study of Ministry of Health, Government of Pakistan, Islamabad, 2002.
314
Available at http://www.crcp.org.pk/PDF Files/foi_moh_study.PDF
315
10. Ali, M., Horikoshi, Y. (2002). Situation analysis of health management information
316
system in Pakistan. Pakistan Health Services Academy, Maternal and Child Health
317
Project Pakistan Institute of Medical Sciences, Islamabad, Pakistan Journal Medical
318
Research 2001; Vol. 41 No. 2. 64-69
319
11. Khan, M. M., Khan, S. H., Walker, D. Fox-Rushby, J., Cutts, F., Akramuzzaman, S.
320
M. Cost of delivering child immunization services in urban Bangladesh: a study
14
321
based on facility-level surveys. Journal of Health, Population and Nutrition 2004;
322
22(4), 404-12.
323
324
325
326
12. Government of India. Annual Report to the People on Health. Ministry of Health and
Family Welfare, Government of India, New Delhi 2010.
13. Green A. An introduction to health planning in developing countries. Oxford: Oxford
University Press 1992.
15