Content uploaded by Jane Eithne Doherty
Author content
All content in this area was uploaded by Jane Eithne Doherty
Content may be subject to copyright.
1
Disease Control Priorities Project
Working Paper No. 37
December 2004
The Cost-Effectiveness of
Primary Care Services in
Developing Countries: A
Review of the International
Literature
Jane Doherty
1
Riona Govender
2
1
Independent consultant, Johannesburg, South Africa and School of Public Health, University of
the Witwatersrand, South Africa.
2
Research Analyst, Health and Development Africa, South Africa.
The Disease Control Priorities Project is a joint effort of The World Bank, the Fogarty International Center of
the National Institutes of Health, the Bill & Melinda Gates Foundation, and the World Health Organization.
2
3
Acknowledgements
We would like to thank Prof. Stephen Tollman of the Wits University School of Public
Health for useful insights during the conceptualisation of this review. Dr. Don de
Savigny of the Swiss Tropical Institute, and Dr. Virginia Wiseman of the London School
of Hygiene and Tropical Medicine, kindly acted as reviewers, providing very useful
comments.
Thanks are also due to: the National Institutes for Health (United States) which funded
Ms. Riona Govender as well as some overhead costs; the Health Policy Unit of the
London School of Hygiene and Tropical Medicine (United Kingdom) which made
available office space and a computer, and organised access to the library, for Dr. Jane
Doherty during a short visit to London; and the Wits University School of Public Health
(South Africa) for facilitating access to the Health Sciences Faculty Library and Centre
for Health Policy Resource Centre.
4
5
TABLE OF CONTENTS
List of acronyms ........................................................................................................6
1. BACKGROUND ..................................................................................................7
2. THE CONCEPT OF GENERAL PRIMARY CARE...............................................7
3. METHODOLOGY ..............................................................................................10
4. WHY ARE PRIMARY CARE SERVICES IMPORTANT? ..................................10
4.1 THE IMPACT OF PRIMARY CARE ON THE BURDEN OF ILL-HEALTH ......10
4.2 THE COST-EFFECTIVENESS OF PRIMARY CARE INTERVENTIONS.......15
4.3 THE AFFORDABILITY OF PRIMARY CARE SERVICES ..............................16
4.4 THE NON-HEALTH IMPACTS OF PRIMARY CARE.....................................19
5. THE EVOLUTION OF THE MINIMUM PACKAGE CONCEPT..........................19
5.1 COMPREHENSIVE VS SELECTIVE PRIMARY HEALTH CARE...................19
5.2 ‘INVESTING IN HEALTH’ ..............................................................................21
5.3 ‘BETTER HEALTH IN AFRICA’......................................................................23
5.4 ‘THE WORLD HEALTH REPORT 2000’ ........................................................26
5.5 THE COMMISSION ON MACROECONOMICS AND HEALTH......................26
5.6 ‘THE WORLD HEALTH REPORT 2002’ ........................................................29
6. THE CRITIQUE OF THE CONCEPT OF COST-EFFECTIVE PACKAGES .......31
6.1 THE METHODS FOR CALCULATING THE BURDEN OF DISEASE.............31
6.2 THE METHODS FOR ESTIMATING COST-EFFECTIVENESS.....................32
6.3 THE USE OF COST-EFFECTIVENESS FINDINGS TO DETERMINE
SERVICE PRIORITIES...........................................................................................33
6.4 ACHIEVING EQUITY.....................................................................................36
6.5 IMPLEMENTING PRIMARY CARE SERVICES.............................................37
7. THE IMPLEMENTATION OF MINIMUM PACKAGES: 3 CASE STUDIES.......38
7.1 THE BAMAKO INITIATIVE: ...........................................................................39
7.2 BANGLADESH ..............................................................................................41
7.3 TEHIP............................................................................................................42
8. DEVELOPING A PRIMARY CARE PACKAGE ON THE BASIS OF ‘GOOD
PRACTICE’: THE CASE OF THE ‘NEED NORMS PROJECT’ IN SOUTH AFRICA 44
9. MOBILISING RESOURCES FOR BASIC PACKAGES.....................................47
10. CONCLUSION: HOW CAN GOVERNMENTS USE INTERNATIONAL
EVIDENCE ON COST-EFFECTIVENESS TO DEVELOP PRIMARY CARE
SERVICES? ..............................................................................................................51
REFERENCES..............................................................................................................57
APPENDIX 1: FIGURES USED TO CONVERT PRICES QUOTED IN REVIEWED
ARTICLES TO 2001 PRICES ($US) .............................................................................63
APPENDIX 2: ORIGINAL PRICES AND OTHER INFORMATION QUOTED IN
REVIEWED ARTICLES ................................................................................................64
6
List of acronyms
AIDS Acquired immune deficiency syndrome
BCG Vaccine to prevent tuberculosis
CMH Commission for Macroeconomics and Health
DALY Disability-adjusted Life Year
DOTS Directly observed treatment, short course
DPT Diphtheria, pertussis and tetanus vaccine
EBM Evidence-based medicine
EPI Expanded Programme on Immunisation
ESP Essential service package
GDP Gross Domestic Product
HAART Highly active anti-retroviral therapy
Hib Haemophilus influenzae type b
HIV Human immunodeficiency virus
IEC Information, education and communication
IMCI Integrated Management of Childhood Illness
ITN Insecticide-treated net
MDG Millennium Development Goal
OI Opportunistic infection
STI Sexually transmitted infection
SWAp Sector-wide approach
TEHIP Tanzania Essential Health Interventions Project
TB Tuberculosis
Tx Treatment
UNICEF United Nations Children’s Fund
US United States
VCT Voluntary Counselling and Testing
WHO World Health Organisation
7
THE COST-EFFECTIVENESS OF GENERAL PRIMARY CARE
SERVICES IN DEVELOPING COUNTRIES:
A REVIEW OF THE INTERNATIONAL LITERATURE
1. BACKGROUND
This review was commissioned in 2003 by the Disease Control Priorities Project (and
funded by the National Institutes for Health) as a background paper to support the
development of a chapter on General Primary Care (Chapter 53) for the forthcoming
second edition of Disease Control Priorities in Developing Countries. The review was
intended to examine the literature that had emerged since the publication of an earlier
review entitled Cost Effectiveness of Primary Health Care (Drummond and Mills 1987).
2. THE CONCEPT OF GENERAL PRIMARY CARE
In this review, general primary care is taken to mean the comprehensive, integrated and
continuing medical and health management of individuals and families when they first
present to the formal health system. General primary care services are taken to include
those delivered at ‘primary level’ facilities (such as health posts, clinics, community
health centres and outpatient departments in district hospitals) and outreach services
(such as mobile services and home visits).
The term ‘General Primary Care’ is distinguished from the ‘Primary Health Care
Approach’ which is a much broader concept developed by the Alma Ata Declaration on
Primary Health Care in 1978 (WHO/UNICEF 1978). The Primary Health Care Approach
is a philosophy that, in some countries, forms the underpinning of the entire health
system, and includes the concepts of equity, public health programmes, decentralisation,
community participation, social development and inter-sectoral action. As a philosophy it
affects the entire health system and ‘implies a re-ordering of priorities that should
permeate all levels and sectors concerned with the promotion of health’ (WHO/UNICEF
(1981), quoted in Tatar and Tatar (1997)). This review examines only that aspect of the
Primary Health Care Approach that relates to individual- and family-based interventions,
although other aspects are considered no less important to the development of healthy
societies.
As part of the Primary Health Care Approach, general primary care should be linked
closely with the needs of the community and integrated with other sectors involved in
community development. It should also be integrated with other levels of care. In fact,
one of the difficulties faced by this review is distinguishing those activities that occur at
the primary level from those that occur elsewhere. In theory, the continuous care of an
individual should form part of a seamless continuum that stretches, as needed, across
all levels of care. For example, in Diagramme 1 interventions for health condition A
stretch across all levels of care, with most activities occurring at the primary level (this
might be the case for pregnancy care, for example). By way of contrast, activities
8
associated with interventions for condition B are mainly located in the community,
with some support being provided by smaller clinics (this might be the case for home-
based care for chronic conditions, for example). Ideally, conditions should be managed
at the type of facility that is most cost-effective. As this review will show, this is often the
primary level. Ideally, then, services should be moved ‘upstream’ – towards the primary
level – as much as possible (in Diagramme 1, see the shifting of the solid black lines
towards the right).
Diagramme 1: The continuum of individual health care
LEVEL OF CARE
Household and outreach Outposts/clinics Community health centres District hospital Regional hosp. Central
Note: The grey shaded areas represent the extent of the activities associated with each
intervention along the continuum of care.
As this review will show, many factors affect which package of interventions is both
needed and cost-effective at a primary level; other factors affect the ability of the health
system to deliver the package at this level adequately (the main factors are summarised
in Box 1). These factors are discussed further in the pages that follow, but the
implication for local health services is that the array of interventions that is appropriate to
meet community needs varies from context to context, as does the ‘cut-off’ between
various levels of care. In practice, limited resources and poor capacity at the primary
level often mean that primary level services perform relatively limited functions (thus, in
Diagramme 1, the solid black line between community health centres and district
hospitals is shifted to the left).
B
C
D
A
9
Box 1: Examples of factors that affect the mix of interventions that is cost-
effective at the primary level
Given the dynamic nature of the relationship between cost-effectiveness, the burden of
ill-health and health system performance, governments and local health service planners
are sometimes hard-pressed to understand how international evidence on the cost-
effectiveness of primary care interventions can be used to plan services on the ground.
It is this critical issue that the review attempts to address in the context of an
understanding that, whereas demands on the primary level are increasing, especially
with the advent of the HIV/AIDS epidemic, the gains made by primary care services over
the past decade have been mixed.
Examples of factors affecting the cost
-
effectiveness of i
nterventions
determined internationally
International issues
• Changes in the global burden of ill-health due to the natural history of disease
• The impact of interventions on the burden of disease
• The methodology used by economic evaluations (e.g. perspective of studies, completeness of data,
the incorporation of time considerations)
Local issues
• Local social, economic and cultural issues
• The extent to which the local burden of disease reflects priorities as identified by global estimates
• The extent to which services are integrated
• The efficiency of service delivery and the extent to which services have already been scaled up
• The role of the private sector and sources of finance for the package (interventions may not be cost-
effective for patients although economic evaluations may determine them to be cost-effective for
providers)
Examples of factors affecting the ability of the primary level to deliver cost-effective packages
International issues
• The development of new technologies
Local issues
• The historical structure and health practice traditions of the health system
• Physical infrastructure
• Geographic, financial and cultural accessibility
• Financial resources
• Human resources and capacity
• The ability to attract and retain staff
• Management and logistical support
• The functionality of the referral system
10
3. METHODOLOGY
Unlike the review published by Mills and Drummond (1987), this review does not attempt
to review individual interventions at the primary level. This is because cost-effectiveness
studies have burgeoned since the early 1990s: many of these studies will be reviewed
(far more expertly than is possible by these authors) for several other chapters in the
second edition of Disease Control Priorities in Developing Countries. Instead, this
review concentrates of the evidence with respect to clusters or packages of basic
primary care services.
The PubMed and International Bibliography of Social Sciences databases were
searched using keywords such as ‘primary care,’ ‘primary health care, ‘evaluation,’
‘costs,’ ‘cost-effectiveness,’ and ‘effectiveness.’ Articles identified by this search were
extracted from the libraries at Wits University in South Africa and the London School of
Hygiene and Tropical Medicine in the United Kingdom, or downloaded from the internet.
Relevant references in articles that had been extracted were also followed up. Some
key informants were contacted for suggestions of other literature that might be useful.
Time and resource constraints meant that the search could not be exhaustive, while
several articles that were identified could not be accessed. An undoubted limitation of
this review is also the shortage of grey literature, especially government reports on the
make-up and experience of country-specific essential health packages. In addition, the
preponderance of articles relate to the African experience. The approach in writing up
the review has therefore been to ensure, first, that a set of core articles has been
collected and, second, that articles represent a range of issues and debates.
All costs quoted in the review are in US dollars and have been converted to 2001 prices,
using inflators calculated from figures supplied by Dr. Jo Mulligan of the London School
of Hygiene and Tropical Medicine in the United Kingdom (see Appendix 1). The original
prices quoted in reviewed articles appear in Appendix 2.
4. WHY ARE PRIMARY CARE SERVICES IMPORTANT?
Before reviewing the evolution of primary care package proposals, it is useful to reflect
on why the primary level, as opposed to other levels, is pivotal to the creation of a well-
functioning health system. Whereas it could be argued that highly cost-effective
interventions deserve to be implemented, no matter the level for which they are
designed, there are unique reasons why those available at the primary level should
receive priority.
4.1 THE IMPACT OF PRIMARY CARE ON THE BURDEN OF ILL-HEALTH
4.1.1 The theoretical argument
Interventions at the primary care level are able to deal, at least theoretically, with 90
percent of health care demands (World Bank 1994:56). Only 10 percent of demands
require the services and skills typically associated with hospitals. In addition, primary
11
care services have the advantage over hospital care that they are more accessible
to the community. Because of their staffing and organisation, they are less costly, and
more easily able to provide comprehensive, integrated, personalised and continuous
care (see Box 2). This argument makes the case for providing services at the primary
rather than hospital level wherever possible, although it does not look at the extent to
which primary care is able to meet health care needs (as opposed to demands) and
reduce significantly the burden of ill-health faced by communities. However, figures are
not available for the total burden of ill-health that could be averted by interventions
based at the primary level. To understand the actual health impact of primary care it is
necessary to look at the empirical evidence.
Box 2: The comparative advantages of health centres
Source: Adapted from World Bank (1994:58)
4.1.2 The empirical evidence
While common sense would predict that primary care has a substantial impact on health
status, this impact is infamously difficult to measure. The key problem is to demonstrate
the causal link between intervention and impact, and especially to distinguish between
the contributions to improved health status of improvements in socio-economic
conditions as opposed to health service delivery. Some authors argue that the impact of
improvements in socio-economic status far outweigh any improvement in health care
delivery (for example, Filmer, Hammer and Pritchett 1997, and Navarro 2000), although
more recently the argument that well-functioning health services can improve health
status, even in the context of poverty, has been made quite strongly (Jha and Mills
2002). Filmer, Hammer and Pritchett (1997) also point to some primary health care
success stories –such as in Kerala (India) and Shanghai (China) – where some health
status indicators are very much lower than might be expected given prevailing
socioeconomic conditions.
Nonetheless, there is surprisingly limited evidence of the share of health improvement
that can be attributed to primary care, even in industrialised countries (Filmer, Hammer
and Pritchett 1997, Starfield 1998). In commenting during the mid-1990s on the
evidence with respect to child mortality reduction in Sub-Saharan Africa, Magnani et al.
(1996: 568) summarise Ewbank and Gribble (1993) as having said that ‘while the
national health programmes of most countries include interventions that have been
shown to reduce mortality in small test programmes, few strong statements could be
made about the overall effectiveness of the large-scale, primary health care programme
Most health problems can be treated with the technology and competence available to well-
functioning health centres.
In 80 to 90 percent of preventive work and for most curative cases, the health centre can
outperform hospitals in terms of continuity, comprehensiveness, integration and cost of
care.
The small scale of the health centre also favours integration of various programmes.
Over-prescription is less common in health centres than in hospitals.
The health centre is more accessible to the community and has the potential to
communicate more ably with the community.
Staff are able to know patients better, which reduces the loss to follow-up of patients.
12
efforts in the region.’ Writing at a similar time, Engelkes (1993:76) comments
cynically that ‘due to a lack of reliable information from the field, donors’ decisions on
PHC have in the past been taken mainly on political grounds.’
Reasons for the lack of evidence lie partly in the nature of the differences between field
trials and routine programmes. These differences are summarized in Box 3. In
addition, it is often difficult to set up an appropriate design to assess the impact of
routine programmes. For example, control areas are difficult to delineate for a number
of reasons, including pressures to provide comprehensive coverage. Also, the detailed
records required for assessment might be prohibitively expensive and, in themselves,
change the nature of the programme (Ewbank 1993). Engelkes (1993) points also to the
severe limitations in primary care projects’ capacity to evaluate themselves, as well as to
the deficiencies in donors’ approaches to evaluation (see Box 4).
3
Box 3: The differences between field trials and routine programmes
Sources: Ewbank (1993), Magnani et al. (1996)
3
Donor-driven evaluations also tend to be performed during the times of year that are least
hostile, thus tending to under-estimate problems (personal communication, Dr. Don de Savigny,
Swiss Tropical Institute).
• Most studies focus on single interventions, whereas actual services tend to combine
interventions;
• Impacts measured in field trials are not necessarily the same as those measured by
routine programmes;
• There is considerable inconsistency between different studies;
• Many studies are located in small geographic areas in a limited number of settings where
generalisability is uncertain, especially if long-term research is being undertaken in the
area;
• The indicators examined in the field trial may not identify and measure the relevant causal
variable;
• The impact of a routine programme may be different because of a different coverage rate,
especially when the programme combines several individually tested interventions (if an
intervention is not delivered to an adequate proportion of the population it will not have the
expected impact); and
• The impact of a routine programme may be different because of differences in the quality
of care provided under field trial versus normal conditions.
13
Box 4: Review of weaknesses in donor-driven evaluations of primary health
care projects
Source: Adapted from Engelkes (1993)
Filmer, Hammer and Pritchett (2000) raise two additional sets of issues that help to
explain why the effects on health improvement of primary care programmes in
developing countries seem, in their view, have been disappointingly small. First, Health
Ministries face enormous difficulties in ensuring that money targeted for primary care is
translated into quality health services on the ground. Bryce et al. (2003), in commenting
on the experience of implementing the Integrated Management of Childhood Illnesses
strategy, provides some concrete examples of such difficulties, including insufficient
training and supervision, high staff turnover, and fragmentation of activities. In fact,
Almeida et al. (2001) contend that historical analyses of the effectiveness of primary
care have tended to ascribe failures to internal weaknesses in the concept, whereas the
external context in which the concept was initiated was fundamentally hostile. They
write that ‘there is no recognition of the cataclysmic effect on public health systems in
less-developed countries of the global economic recession of the 1980s and the
application of policies stressing privatisation and decreased public spending in that
decade and the next, which resulted in rising poverty and under-funding of health
services in many less-developed countries, to the point of near-collapse in the poorest
countries’ (Almeida et al. 2001).
Filmer, Hammer and Pritchett (2000) also contend that the provision of public primary
care services may sometimes have ‘crowded out’ the consumption of equally effective
services rendered by the private sector, which might explain the observed, limited impact
of the curative component of primary care services. This last point highlights the
importance of measuring the net impact of packages. Magnani et al. (1996) provide
another example of where this is important, noting that ‘gains realised through single
interventions might be offset by continued high levels of exposure to other risk factors for
mortality in low-income country settings; that is, through “replacement mortality”.‘
These problems notwithstanding, some studies have been able to demonstrate success
in the area of child health. Bryce et al. (2003) point to a number of successful small-
scale projects. On a larger scale, Magnani et al. (1996) show, using data from a national
survey in Niger, that children living in villages near to health dispensaries were 32
percent less likely to die than children without access to modern primary care services
(differential access was due to the phased implementation of services which represented
Many donors did not have guidelines. If guidelines existed, they were often more
concerned with format than methodology.
Terms of reference for external evaluators were often vague and tended to be donor-
oriented. External consultants were left to collect information in the way they wanted.
Evaluations by external evaluators usually had to be performed within two to three weeks
which is very short when a country is unknown to a consultant and a project has been
running for several years.
Few projects had internal evaluation and monitoring systems and data, if they were
collected, were not collected systematically. External consultants were therefore seldom
confronted with objective information.
National counterparts to foreign consultants were usually passed over or unable to
participate as equals (due to lack of time and information) in evaluations.
14
a natural quasi-experiment). The use of multi-variate and other analyses suggest
that these results were not due to the location of dispensaries in villages that were
predisposed to lower mortality. Magnani et al. (1996:574) conclude that ‘packages of
basic primary health care services can be effectively mounted at the national level so as
to have a significant impact on infant-child mortality over a fairly short period of time.,.’
but acknowledge that the impact of services may have been exaggerated by the high
initial level of mortality, as well as the occurrence of famine and a severe measles
outbreak during the study period. However, these sorts of conditions are not uncommon
in many developing countries.
In a separate study, Ewbank (1993:S64,S71) states that the results of surveys in Zaire
and Liberia ‘suggest that child survival programmes in Africa can reduce mortality
substantially in populations living in different environments at very different initial levels
of child mortality … In both countries, it appears that the programme reduced mortality
under age 5 by about 20% or more.’ The author assesses these findings as robust
because of their constancy under differing circumstances, as well as their consistency
with other studies. In addition, this study improved on earlier surveys because it
examined a longer follow-up period, assessed routine rather than special programmes,
and was not limited simply to immunization activities (including, as it does, oral
rehydration services and treatment with anti-malarial drugs). Ewbank (1993:S71) notes,
however, that ‘it is not clear to what extent each of the programme components
contribute to the reduction of mortality. While the cumulative evidence on measles
vaccination is quite impressive, there is much less evidence on the contribution of BCG,
DPT, home-based use of oral rehydration, and presumptive treatment of fevers with
chloraquine. Therefore, while the overall programme is apparently successful, the
optimal combination of interventions to reduce mortality has yet to be determined.’
Given the paucity of developing country evidence, it is useful to turn to the experience of
the United States. Shi (1994) finds that, in the United States, primary care is ‘by far the
most significant variable related to better health status, correlating to lower overall
mortality, lower death rates due to diseases of the heart and cancer, longer life
expectancy, lower neonatal death rate, and low birth weight.’ In studies undertaken by
Shi and Starfield (2000, 2001) on income inequality and primary care, it was established
that there exists a significant association between higher primary physician supply and
good health status, even in the context of higher income inequality: ‘The findings of a
significant association between primary care and self-rated health contributes to the
mounting evidence that specific aspects of health care services have an independent
effect on improving population health, in particular, the beneficial effects of primary care’
(Shi and Starfield 2000). The authors suggest, therefore, that, within the particular
setting studied, strengthening of the primary care aspects of health services could
mitigate some of the adverse impacts that income inequality has on individuals’ health
status.
In reviewing a number of studies that look at the ‘efficacy’ of primary care services for
vulnerable populations in the United States, Blumenthal, Mort and Edwards (1995) find
considerable evidence of positive impacts, especially on utilisation (see Box 5). They
make the point that the literature does not adequately address the issue of whether
primary care reduces the cost of care for under-served populations but conclude that ‘a
commitment to primary care should be made for its potential to improve the satisfaction
and health status of the American public, not for its potential to save money’
(Blumenthal, Mort and Edwards 1995:269).
15
Box 5: Evidence of the efficacy of primary care services
4.2 THE COST-EFFECTIVENESS OF PRIMARY CARE INTERVENTIONS
Source: Adapted from Blumenthal, Mort and Edwards (1995)
4.2 THE COST-EFFECTIVENESS OF PRIMARY CARE INTERVENTIONS
In 1993, the World Bank published the first detailed figures on the global cost-
effectiveness of different interventions in its World Development Report, Investing in
Health (World Bank 1993).
4
The report found that, in countries with moderate to high
mortality, only a few causes accounted for the majority of the burden of ill-health. In
1990, fifty-five percent of the global burden of disease was concentrated in children
under 15, and 75 percent of this burden was caused by 10 disease conditions or clusters
(Bobadilla et al. 1994). Except for congenital malformations, all these causes
correspond to very cost-effective interventions, most costing less than $100 per
disability-adjusted life year (DALY)
5
averted. Together, it was estimated, these
interventions could eliminate 21 to 28 percent of the burden of ill-health in children.
The burden of adult disease was found to be less concentrated: here, the ten main
causes of disease and injury accounted for only 50 percent of the burden.
6
Most
interventions are quite cost-effective but the impact is moderate because they only
prevent or treat a small fraction of the problems. Such interventions would only
eliminate 10 to 18 percent of adult disease burden.
Amongst the highly cost-effective interventions against both the childhood and adult
burden of ill-health that were identified by Investing in Health as part of a ‘minimum
package of health services,’ were some of the classic components of primary care (see
the ‘clinical services’ listed in Table 1). Indeed, all of the ‘public health’ activities, except
possibly the school health programme, include some element of individual service
delivery in the primary care setting.
4
Strictly speaking, the report depended on cost-utility analysis but the term cost-effectiveness
analysis has become indelibly associated with the report.
5
A DALY is the sum of the burden of disease due to premature death and that due to non-fatal
disability.
6
Bobadilla et al. (1994) comment that separating interventions for age group is artificial as
benefits accrue in later life (e.g. hepatitis vaccine) and improve well-being (e.g. cognitive abilities).
Adult interventions also have benefits for children (HIV prevention, prenatal care).
1. Community-based interventions improve access to services, reduce the use of emergency
and outpatient departments at hospitals, increase the use of non-institutional ambulatory
care, and reduce the use of hospital care (especially with respect to preventable
hospitalisations).
2. Primary care is associated with improved control of routine illnesses that have serious
consequences if untreated.
3. The availability of primary care services improves patients’ self-perceived health status.
4. The longitudinal care afforded by primary care services is independently associated with
improved patient satisfaction, reduced use of ancillary and laboratory tests, improved
patient compliance, shorter length of stay, and improved recognition of patients’
behavioural problems.
16
Table 1: Cost-effectiveness of the health interventions (and clusters of
interventions) included in the minimum package of health services
as recommended by the World Bank Development Report (1993) for
low- and middle-income countries (2001 prices)
COST PER DALY ($) INTERVENTIONS
Low-income
countries
Middle-income
countries
Public health
Expanded programme of immunization plus (i.e. including
vaccine against Hepatitis B and Vitamin A supplementation)
15-22 32-38
School health programme 25-32 48-54
Tobacco and alcohol control programme 44-70 57-70
AIDS prevention programme
**
4-6 16-23
Other public health interventions (includes information,
communication and education on selected risk factors and
health behaviours, plus vector control and disease
surveillance)
* *
Total
18
Clinical services
Chemotherapy against tuberculosis 4-6 6-9
Integrated management of the sick child 38-63 63-127
Family planning 25-38 127-190
STD treatment 1-4 13-19
Prenatal and delivery care 38-63 76-139
Limited care (includes treatment of infection and minor trauma;
for more complicated condition, includes diagnosis, advice and
pain relief, and treatment as resources permit)
253-380 507-760
Total
* 168
Notes:
* information is not available for the cells that have been left blank, presumably because the authors were not
able to aggregate data up to this level
** this understates cost-effectiveness because the analysis looked at the probability of transmission to others only
in the first year
Source: Bobadilla et al. (1994:657)
4.3 THE AFFORDABILITY OF PRIMARY CARE SERVICES
Drummond and Mills (1987) found the best estimate of the cost of effective PHC
(including the recurrent and capital costs of basic and village-level health services but
not of water and sanitation) to be 2 percent of annual per capita GNP. This was based
on the annual per capita costs of demonstration projects that ranged from $2.85 to
$27.26 in 2001 prices (see Table 2).
7
These are considerable amounts, given that many
developing country governments do not spend as much as 2 percent of annual per
capita GNP on their entire health sector.
7
Drummond and Mills (1987) argue that, while the costs of large-scale PHC programmes may be
lower than these estimates, they may well not be providing effective services.
17
Table 2: The costs of six primary health care projects reviewed by Gwatkin et
al. (1989) (2001 prices)
LOCATION PRICE BASE
(yr)
ANNUAL PER
CAPITA COST
(2001 prices)
% OF ANNUAL
PER CAPITA
GNP
Imesi, Nigeria 1966 6.79
2.0
Etimesgut,
Turkey
1968-74 27.26-22.48
1.5-2.0
Narnangwal,
India
1970-73 3.04-6.54
0.8-2.0
Rural Guatemala
II
1969-77 *13.96/8.53
0.75-1.0
Jamkhed, India 1978 2.85-3.41
1.00-1.25
Kavar, Iran 1975 9.59-14.67
0.4-0.5
* Only one price was quoted in the original article: as it was not clear to which year this price applied, a
conversion has been supplied for both 1969 and 1977
Source: Drummond and Mills (1987:77)
The Investing in Health estimates for the minimum package outlined in Table 3 was $15
per capita per annum in low-income countries and $28 in middle-income countries (2001
prices). The difference in costs between low- and middle-income countries are due to
different demographic structures, epidemiological profiles, burdens of disease and
labour and other input costs. Two thirds of the cost - $9.9 and $18.6 in low- and middle-
income countries respectively - was devoted to clinical services. These estimates were
calculated through two methods, one by summing the costs of individual interventions,
the other by costing a ‘prototype’ district health system capable of delivering the package
(Bobadilla et al. 1994). The prototype consisted of a district hospital, health clinics and
outreach activities, and required one district hospital bed per 1000 population, 0.1
physicians per 1000 population, and 2 to 4 nurses per physician. Costs were based on
effective interventions, rather than the costs of actual, often ineffective services on the
ground.
On the whole, the World Bank figures are higher than earlier estimates. In addition,
achieving universal coverage would probably raise marginal costs substantially above
average, because of the added costs of extending service delivery to people living in
remote areas. It is typically these poor that need public subsidies the most. Bobadilla et
al. (1994:661) note that, in these instances, ‘the relative importance of cost-effectiveness
versus equity will then determine whether to modify the package by leaving out some
interventions, providing mobile services rather than fixed facilities, concentrating on
public health rather than clinical interventions for the high-cost population, or sacrificing
some efficiency in order to preserve equity.’ Another justification for accepting a high
marginal cost would be the potential to eradicate a disease completely, as thereafter
there would be no further costs (as is the case for smallpox).
18
Table 3: The per capita costs of delivering the World Bank minimum package
(2001 prices)
ANNUAL COST PER CAPITA
($)
INTERVENTIONS
Low-income
countries
Middle-income
countries
Public health
Expanded programme of immunization plus (i.e. including
vaccine against Hepatitis B and Vitamin A supplementation)
0.6 1.0
School health programme 0.4 0.8
Tobacco and alcohol control programme 0.4 0.4
AIDS prevention programme
2.2 2.5
Other public health interventions (includes information,
communication and education on selected risk factors and
health behaviours, plus vector control and disease
surveillance)
1.8 3.9
Total
5.3 8.7
Clinical services
Chemotherapy against tuberculosis 0.8 0.3
Integrated management of the sick child 2.0 1.4
Family planning 1.1 2.8
STD treatment 0.3 0.4
Prenatal and delivery care 4.8 11.1
Limited care (includes treatment of infection and minor trauma;
for more complicated condition, includes diagnosis, advice and
pain relief, and treatment as resources permit)
0.9 2.7
Total
9.9 18.6
Grand Total
15.0 27.2
Source: Bobadilla et al. (1994:657)
Thus, although the minimum package is able to control large disease burdens and at
little cost compared to a host of other interventions, the fact remains that many poor
countries are unable to finance the minimum package themselves. On average,
governments tend only to spend $7.6 per capita in 2001 prices, a figure which rises to
$17.7 only when private sources are included (Bobadilla et al. 1994). This implies the
need for increased public spending on health, reorientation away from discretionary
services (that is, services not in the minimum package), targeted public spending on the
poor, and the harnessing of private and donor resources. The fact that primary care –
even of the most elementary kind - is presently unaffordable to most poor countries is of
great concern, and lies behind the call by the Commission for Macroeconomics and
Health (Jha and Mills 2002) for massive donor commitments to tackle health problems
in these countries. It is probably also partly responsible for the continued
implementation of vertical programmes, despite continued punting of the ‘minimum
package’ concept, as shown in a later section.
19
4.4 THE NON-HEALTH IMPACTS OF PRIMARY CARE
While most of the recent literature on primary care packages places value on primary
services because of their ability to reduce the burden of disease considerably and at low
cost, there are potentially other benefits that such services bring to society. Some of
these – such as accessibility, continuity and improved communication - have already
been referred to in Box 2. However, amongst the most striking benefits may be the
welfare benefits that accrue to households as a result of the prevention of severe
disease. Severe disease can limit the ability of patients and caregivers to work, and also
lead to the consumption of household assets in the purchasing of care. Russell (2003)
finds that such costs amounted to just over 10 percent of household income in three
developing country settings studied, proportions which can have a catastrophic impact
on the sustainability of poor households. Through prevention and early treatment,
accessible primary care services can reduce the negative economic consequences of ill-
health for households, reduce absenteeism and enhance children’s performance at
school.
Primary level services are also potentially more responsive to patients’ non-health
needs. These include the need for health services to meet community expectations and
to treat patients in a pleasant manner. In addition, primary level services have the ability
to act as community resources (providing meeting places, for example) and to engage in
community development activities.
All in all, well-functioning primary level services represent the ‘face’ of the health system,
and have the potential to inspire trust in the system as a whole. In reality, primary level
services undoubtedly often fall far short of this potential. However, this does not explain
the scant attention paid to these features in analyses of the ‘value for money’ provided
by this level of care.
5. THE EVOLUTION OF THE MINIMUM PACKAGE CONCEPT
5.1 COMPREHENSIVE VS SELECTIVE PRIMARY HEALTH CARE
The 1978 Declaration of Alma-Ata focused international health care efforts on low-cost,
low-complexity interventions of both the medical and social sort, and at both the primary
and community levels. In particular, the Declaration emphasized the importance of
primary health care as a strategy for transforming the health system and contributing to
community development through, for example, community involvement in decision-
making and the supply of water and sanitation. The health care interventions were
described only in broad terms and were not costed. Indeed, the recommendations were
based on very incomplete data on the burden of disease, especially with respect to non-
communicable diseases and the causes of disability, and health gains at this time were
largely measured in terms of the reduction of mortality (World Bank 1994). Cost-
effectiveness data were also very limited. Nonetheless, the suggested health care
interventions were very similar to what were later proposed by the World Bank. Overall,
though, the Alma-Ata conceptualisation of primary health care is perceived to have been
much more comprehensive and patient-centred.
20
Soon after the Alma Ata Declaration, the apparent unaffordability of the Primary
Health Care Approach led to the emergence of the concept of ‘selective primary health
care’ as expressed, for example, by Walsh and Warren (1979). This advocated focusing
initially on a limited number of components, often directed towards children’s health and
individual tropical diseases. Selective interventions were often centrally planned, and
managed and operated by dedicated staff. They were intended as entry points into the
health care system, and proved especially useful for eradication campaigns and dealing
with epidemics following natural disasters (World Bank 1994).
However, the selective approach was criticized for not acknowledging that primary care
must take account of the range of diseases that present, some of which were not
included in the selective agenda (World Bank 1994). In addition, administrators
implementing what were essentially vertical programmes often have little contact with
local officers, and seldom co-ordinate well with other vertical programmes. Briggs,
Capdegelle and Garner (2003) note that fragmentation may lead to duplication of
training, supervision and logistics management. Other inefficiencies are caused by the
need for specialised staff which in turn leads to greater numbers of staff, and the wasting
of service users’ time if they have need of multiple services. ‘Verticalisation’ can also
lead to competition between programmes, the favouring of some issues at the expense
of others, poor continuity of care, the disruption of routine health care and the disruption
of national capacity.
In commenting in 1993 on progress since the Alma-Ata Declaration, Engelkes (1993:72)
writes that ‘more than 10 years have elapsed since then, and initial optimism has been
replaced by scepticism, criticism and, in many instances, a change of direction from
horizontal, integrated, comprehensive PHC projects towards vertical, selective activities
like immunisation campaigns.’ Writing at a similar time, the World Bank (1994:50) notes
that ‘in the years subsequent to the Alma-Ata Declaration, efforts to provide primary
health care have taken the form of either highly selective, ‘vertical’ programmes
designed to deal separately with specific health problems, or broader proprammes
involving community or village health workers. In most African countries, neither of
these strategies has done much to persuade policymakers to shift resources away from
curative care to a well-defined package of cost-effective primary and preventive care
services.’
This sense of disillusionment with vertical programmes – together with the continued
misallocation of resources towards expensive, cost-ineffective care - led to the
development of the concept of ‘packaging’ of services in the early 1990s, culminating in
Investing in Health. This tendency not withstanding, vertical services still continue in
many instances, reflecting the constraints faced by developing countries in mounting a
full set of services. Vertical programmes are also attractive to donors, as it is easier to
set targets and monitor results for such programmes (Unger, de Paepe and Green
2003).
It has to be acknowledged that, in some instances, centrally planned and vertically
delivered services may perform better than services which are locally planned and
delivered and integrated with existing services (Unger, de Paepe and Green 2003).
Briggs, Capdegelle and Garner (2003) note that there is as yet no concrete evidence to
21
suggest that decentralised or ‘horizontal’
8
services are superior to vertical services
in terms of impact. This conclusion was based on the four sole studies that have set out
to compare the relative benefits of horizontal versus vertical services, and the authors
identify methodological flaws in each of these, so there is clearly a need for further
evaluations. However, there is some evidence that horizontal services are less costly,
as one would have expected from the inefficiencies associated with verticalisation. This
has important implications for the sustainability of health services. Unger, de Paepe and
Green (2003) make a strong theoretical argument for the desirability of decentralised
services in terms of the practical management of patients.
A recent addition to the horizontal-vertical debate has been the move towards clustering
of services into ‘integrated programmes’ that attempt to do away with the excessive
separation implied by vertical services, whilst retaining some sort of focus, the
Integrated Management of Childhood Illness (IMCI) being one such example.
Integration bundles services across several diseases using a common delivery
technology and point of contact with the beneficiary. In doing so, it addresses more of
the burden of disease at less cost than would individual interventions separately,
improving efficiency. It also facilitates the training of health workers in dedicated skills.
An integrated intervention can be delivered in either a vertical manner (as once was
done for EPI) or in a decentralized manner, as frequently done now.
5.2 ‘INVESTING IN HEALTH’
Investing in Health was premised on the assumption that no country can afford to
provide all the health care services that are needed by its population, and that it is
therefore important to establish criteria for which services will be funded. The two most
important criteria, as identified by the report, are the size (or potential size) of the burden
caused by a disease, injury or risk factor, and the cost-effectiveness of interventions to
deal with it. Only if an intervention is cost-effective, while simultaneously eliminating a
large proportion of the burden of disease, should it become a priority.
The importance of the first criterion led to the establishment of the Global Burden of
Disease enterprise of the World Bank and WHO. This enterprise has three central aims
(Murray and Lopez 2000:70):
(i) ‘to decouple epidemiological assessment of the magnitude of health
problems from advocacy by interest groups of particular health policies or
interventions;
(ii) to include in international health policy debates information on non-fatal
health outcomes along with information on mortality; and
(iii) to undertake the quantification of health problems in units that can be used in
economic appraisal’
A second undertaking was the review of cost-effectiveness data to identify a minimum
package of interventions able to address the main health problems. While the concept
8
Briggs, Capdegelle and Garner (2003), and Unger, de Paepe and Green (2003) use the term
‘integrated’ rather than ‘horizontal/decentralised’ However, in this document, the term ‘integrated’
is used in a different sense.
22
of cost-effectiveness was a strong thrust of the report, so was the concept that
interventions should be packaged together. This represented a shift away from the
tendency to verticalise health programmes (although it did not go as far as to endorse
the philosophy of comprehensiveness put forward by the Alma-Ata Declaration). The
justification of the need for packaging of services was as follows (Bobadilla et al. 1994):
1. When governments simply fund or provide a list of services without considering
their relationship, they do not take account of either the joint costs of
programmes or the co-morbidities experienced by patients.
2. When governments simply pay for a collection of inputs and leave decisions on
services to health care workers, or to patients who demand services, then
services of questionable value tend to be rendered.
3. Thus, ‘the principal argument for a collection of services to be provided jointly is
to minimize the total cost of the package by exploiting the shared use of inputs
and by reducing the cost to patients of obtaining services. Clustering of
interventions improves cost-effectiveness through at least three mechanisms:
synergism between treatment or prevention activities; joint production costs; and
improved use of specialized resources through the screening of patients at the
first level of care, to ensure that a small share of high-risk cased can be
recognized and referred to hospital. Sometimes a cluster of diseases can be
treated together, because they share diagnostic procedures or treatment
protocols, or even the same drugs. And sometimes services can be organised to
reach related individuals e.g. integration of maternal and child care. Thus the
package becomes more than simply a list of interventions: properly understood,
it is also a vehicle for orienting demand and improving referral’ (Bobadilla et al.
1994:654).
4. Another, non-medical reason for packaging is that governments tend to find it
difficult to set priorities and plan investments, and a minimum package – as well
as an essential package (which, in the terminology of Investing in Health means
any additional priority services that can be afforded above the minimum), creates
a useful basis for planning.
The cost-effectiveness estimates used by Investing in Health were derived from Jamison
et al. (1993), with some modification. As indicated earlier, costs were calculated per
disability-adjusted life-year (or DALY). In turn, DALYs were calculated as the sum of
losses to premature mortality and disability. Because of methodological difficulties
experienced at this time in estimating the burden of disease due to different risk factors,
only a few interventions against risk factors could be evaluated. Cost-effective
interventions against those factors that resulted in a large burden of disease – or had the
potential to cause a large burden of disease (such as the growing HIV/AIDS epidemic
and increasing tobacco consumption) – were included in the package. A notable
exception was water and sanitation which is cost-ineffective according to economic
evaluations which do not consider non-health benefits.
The cost-effectiveness of interventions was seen to vary greatly and ‘by much more than
the likely errors of estimation or the variation in cost-effectiveness from one country or
epidemiological situation to another’ (Bobadilla et al. 1994:654). This, together with the
fact that some of the most cost-effective interventions were ones that dealt with large
23
burdens of disease,
9
made it relatively easy to distinguish which services should be
included in a minimum package and which should not. Bobadilla et al. (1994:654) were
therefore able to conclude that ‘it matters which services are included in a package; this
would not be the case if the cost per healthy life-year gained were about the same for all
services.’
5.3 ‘BETTER HEALTH IN AFRICA’
Investing in Health was followed in 1994 by another World Bank publication, Better
Health in Africa (World Bank 1994). This publication drew on the thinking of the 1993
report but applied it more specifically to the African continent. In particular, it attempted
to give guidance on how to operationalise ‘a cost-effective approach to health’ (World
Bank 1994:4). It identified three ‘underpinnings’ for such an approach, the first of which
is the use of cost-effective packages of services targeted at the main health problems.
The second underpinning is the decentralization of service delivery to the district level,
including the expansion of health centres and district hospitals.
The report was based on the assumption that ‘systems composed of well-functioning
health centres and first-referral hospitals are capable of responding to, and
accommodating, more than 90 percent of health demands in an average rural/peri-urban
district’ (World Bank 1994:129). The report made explicit the types of services that
should be available at each level of facility (see Table 3 in Appendix 2), perhaps
correcting the impression given by a superficial reading of Investing in Health that the
basic package is delivered only at the primary level. The third underpinning is the
improved management of personnel, pharmaceuticals and infrastructure.
The Better Health in Africa package included more or less all the services proposed by
Investing in Health (with both public health and clinical services being termed ‘individual
health services’), although it is generally difficult to compare package proposals as the
level of detail provided on each intervention tends to vary considerably (see Table 4). A
notable exclusion in Better Health in Africa is malaria prevention. A notable inclusion,
however, is water and sanitation, something that had not been part of the health
package priorities put forward by Investing in Health, for the reasons mentioned earlier.
It is only in 2002 that water-related interventions (but not sanitation) make their way back
onto cost-effectiveness lists.
9
Cost-effective interventions were not included for conditions that are very rare, or which result in
negligible individual health loss. If a rare condition causes large losses to individuals it was seen
as a candidate for inclusion in the expanded version of the package (i.e. the essential package).
24
Table 4: A comparison of the different basic packages proposed by WHO/UNICEF, the World Bank, WHO and the
Commision on Macroeconomics and Health
INTERVENTION
Alma Ata
Declaration
(1978)
Investing in
Health
(1993)
Better Health in
Africa
(1994)
WHO report
2000
CMH Working
Group 5
(2002)
WHO report
2002
Maternity-related interventions
+ + + + +
Antenatal care + + + +
Treatment of complications during pregnancy + + + +
Skilled birth attendance + + + +
Emergency obstetric care + + + +
Postpartum care + + + +
Family planning + + + +
Nutrition: pregnant and lactating women +
Tetanus toxoid +
Childhood disease-related interventions (prevention):
+ + + + + +
BCG + + + + +
Polio vaccine + + + + +
DPT + + + + +
Measles + + + + +
Hepatitis B + + + +
Hib + + +
Vitamin A supplementation + + + + +
Iodine supplementation + + + +
Zinc supplementation +
Anthelmintic treatment +
School health programme (incorporating micronutrient supplementation,
school meals, anthelmintic treatment, health education)
+ + +
Childhood disease-related interventions (treatment):
+ + (as part of
IMCI):
+ + (as part of
IMCI)
+ +
Acute respiratory infections + + +
Diarrhoea + + + +
Causes of fever + + +
Malnutrition + (inlc. nutrition
& suppl. feeding)
+ +
Anaemia + +
Feeding/breastfeeding counselling +
Malaria prevention
* + + +
Insecticide-treated nets + +
Residual indoor spraying +
Malaria treatment
* + +
Tuberculosis treatment
* + + + +
DOTS for smear positive patients + +
25
DOTS for smear negative patients +
HIV/AIDS prevention
+ (more ltd. than
later packages?)
+ + +
Youth focused interventions +
Interventions working with sex workers and clients + +
Condom social marketing and distribution +
Workplace interventions +
Strengthening of blood transfusions systems + +
Voluntary counselling and testing + +
Prevention of mother-to-child transmission +
Mass media campaigns + +
Treatment for sexually transmitted diseases + + + +
HIV/AIDS care
+ +
Palliative care + (see under
limited care)
+
Clinical management of opportunistic illnesses + +
Prevention of opportunistic illnesses +
Home-based care +
HIV/AIDS HAART provision
+
Tobacco control programme (taxes, legal action, information, nicotine
replacement)
+ + +
Alcohol control programme
+
Other public health interventions (includes information, communication
and education on selected risk factors and health behaviours, plus vector
control and disease surveillance)
+ + + (IEC)
Limited care (includes treatment of infection and minor trauma; for more
complicated conditions includes diagnosis, advice and pain relief, and
treatment as resources permit)
+ + +
Non-communicable diseases and injuries (selected early screening
and prevention)
+ +
Population-wide interventions to reduce the risks of cardiovascular
disease (salt and cholesterol lowering strategies)
+
Water and sanitation
+ + +
(disinfection
at point of
use)
Sources: World Health Organisation and UNICEF (1978), World Bank (1993), World Bank (1994), World Health Organisation (2000), Commission for Macroeconomics and Health
(2001), World Health Organisation (2002)
** This report only addresses interventions against risk factors.
* These, and other disease prevention and control initiatives, fell under a general item termed ‘prevention and control of locally endemic diseases’ (HIV/AIDS was not an issue
at the time)
Note: Where a ‘+’ appears in a grey shaded area but not in the white cells beneath this area, this means that no details of the exact interventions were provided in the report.
26
Another interesting feature of this particular package is that it includes the institutional
support necessary to achieve both individual health care as well as inter-sectoral
interventions. This item included national management support, training, the
development of district teams, and a financial incentive to retain staff in peripheral
services. All in all, then, the package proposed by Better Health in Africa has more of
the character of a plan for implementing cost-effective district health services than the
more narrowly described package in Investing in Health.
The Better Health in Africa package was estimated to cost, in 2001 prices, roughly $17
per capita per annum in low-income African countries, primarily in rural and peri-urban
settings (this is slightly higher than the Investing in Health estimate). Fifty-nine percent
($9.81) of this was devoted to personal health care, 30 percent ($5.04) to inter-sectoral
interventions, and 11 percent ($1.90) to supporting services. It was estimated that in
higher-income African countries, such as Zimbabwe, the package could cost 20 to 25
percent more, up to roughly $20 per capita per annum.
10
These estimates were based
on the costs of well-functioning facilities and inter-sectoral programmes in several
African countries, and included both recurrent and annualized costs of capital
investments.
5.4 ‘THE WORLD HEALTH REPORT 2000’
The World Health Report 2000, published by the World Health Organisation, listed a
number of interventions that are cost-effective (see Table 5 in Appendix 2), but did not
put a cost of this ‘package,’ or elaborate on how the methodology used to derive this list
differed from the World Bank approach. The list expands on earlier lists by widening
considerably the range of HIV/AIDS-related prevention activities (reflecting improved
knowledge of the range of cost-effective interventions) as well as explicitly mentioning
non-communicable disease screening and prevention (reflecting the growing awareness
of the health transition being experienced in developing countries) (see Table 4). It is
difficult to tell whether other discrepancies with earlier proposals reflect conscious
decisions or simply differences in the level of detail contained within different proposals.
5.5 THE COMMISSION ON MACROECONOMICS AND HEALTH
In 2002, Working Group 5 of the Commission on Macroeconomics and Health published
a report on its deliberations, entitled Improving Health Outcomes of the Poor (Jha and
Mills 2002). The report reviewed the health sector interventions that could do most for
the health of the poorest billion of the world’s population (those living in all the countries
of Sub-Saharan Africa and all other countries with a per-capita GNP below $1,200 (1999
prices)),
11
and estimated the benefits and costs of increased provision of these
interventions. The interventions reflect earlier thinking on packages, although HIV/AIDS-
10
This is because of the higher intensity of demand that results partly from higher levels of
household education, reinforced by higher income levels.
11
In 2001 prices these would be countries below $1,255.
27
related interventions became more detailed. As with earlier packages, the report
proposed that interventions be delivered through outreach services, health centres and
local hospitals, which were referred to as ‘close-to-client’ health services to avoid the
ambivalences present in other terminology.
The approach to deriving this list of interventions clearly differed from that used in earlier
proposals for packages. First, there were modifications in the method for calculating the
burden of disease: premature mortality alone was used and target life expectancies
were lowered. Mortality was seen by the Commission as a key indicator for three
reasons (Jha and Mills 2002). First, it frequently accounts for the bulk (roughly three-
quarters) of the burden of ill-health as measured by more sophisticated approaches.
Second, morbidity correlates reasonably closely with mortality in the conditions identified
as the main problems (so measures to reduce mortality will tend to reduce morbidity).
Third, data on mortality tend to be more reliable than data on disability. The
Commission was at pains to note that the use of mortality is not meant to minimize the
importance of injury and mental disorders (where there can be significant morbidity
without mortality, and for which very cost-effective interventions are emerging).
With respect to cost-effectiveness judgements, these were based on a range of reviews
that were themselves reviewed. Interventions were chosen if they were effective,
capable of being scaled up relatively quickly and applied widely, and responsive to
important epidemiological issues (see Table 6 in Appendix 2). As in earlier
calculations, however, non-health benefits were not considered.
Costs were estimated for scaling up 49 priority health interventions to target levels for
2015 in 83 poor countries. Apart from the recurrent and capital costs of the
interventions themselves, four additional sets of expenditures were included. These
were management costs generated at levels above the ‘close-to-client’ services,
expenditure to improve absorptive capacity, expenditure on the improvement of the
quality of care, and 100 percent increases in staff salaries to deal with the problems of
staff recruitment and retention. It was estimated that an additional $40 to $52 billion
annual expenditure would be required by 2015 to scale up the selected interventions to
reach high levels of global coverage (and consequently achieve high levels of benefit).
In 2001 prices, this would bring total per capita expenditure – including the public and
private sectors - to $41 in all countries (but $40 in the poorest countries), from a base in
2002 of $25 in all countries (but $13 in the poorest) (see Table 5 (the original 2002
prices appear in Table 7 of Appendix 2)).
Table 5: Baseline annual per capita expenditure in 83 poor countries in 2002,
and total projected expenditure to achieve 2015 targets (2001 prices)
COUNTRY
GROUPING
2002 BASELINE EXPENDITURE TOTAL EXPENDITURE IN 2015
Least developed 13
40
Other low-income 24
36
Lower middle-income 27
39
Upper middle-income 260
331
All countries 25
41
Source: Commission for Macroeconomics and Health (2001)
28
Table 6: A comparison of the total annual per capita costs of minimum
packages calculated by different reports (2001 prices)
REPORT Low-income countries Middle-income countries
Investing in Health (1993)
15 27
Better Health in Africa (1994)
16-20
Commission for
Macroeconomics and Health
(2001)
Least developed:
40
Other low-income:
36
Lower middle-income:
39
Upper middle-income:
331
Table 6 compares the cost estimates of the Commission for Macroeconomics and
Health with those of the earlier reports, Investing in Health and Better Health in Africa.
The most recent data suggest that, in low-income countries, a minimum package could
cost around twice the amount estimated by the earlier reports. This is probably due, in
the main, to the acknowledged need to extensively upgrade the capacity and quality of
district health systems to deliver the package effectively. The fact that even the pared
down packages of the mid-1990s are presently not affordable to most low-income
countries, highlights the need for considerable aid to be channelled to such countries. It
also highlights the importance of finding ever more efficient strategies for delivering
these packages to populations, an issue which is dealt with in some degree in a later
section. Interestingly, the report of Working Group 5 of the Commission for
Macroeconomics and Health provided some examples of how interventions could be
structured between levels of care (see Table 7). Here it was considerably more specific
than Better Health in Africa, providing clearer guidance to country governments on how
to organise the delivery of cost-effective interventions.
29
Table 7: Examples of intervention delivery by level of care, as proposed by
the Commission for Macroeconomics and Health
Level
of care
TB Malaria HIV/AIDS Childhood
diseases
Maternal/
perinatal
Smoking
Hospital DOTS for
complicated
TB cases
Tx of
complicated
malaria
Blood transfusion
for HIV/AIDS;
HAART Tx of
severe OI for
AIDS;
Palliative care
IMCI: severe
cases
Emergency
obstetric care
Health
centre or
post
DOTS Tx
uncomplicated
malaria;
Intermittent Tx of
pregnant women
for malaria
Anti-retrovirals plus
breast milk
substitutes for
mother-to-child
transmission;
Prevention of OI,
and Rx of
uncomplicated OI;
VCT;
Tx of STIs
IMCI;
Immunisation;
Tx of severe
anaemia
Skilled birth
attendance;
Antenatal and
postnatal care;
Family
planning post
partum
Cessation
advice;
Pharmacol-
ogical
therapies
for smoking
Outreach
services
Epidemic
planning and
response
Peer education for
vulnerable groups
Specific
immunization
campaigns;
Outreach IMCI:
home
management of
fever;
Outreach for
micronutrients
and deworming
Outside
health
sector or
not
involving
direct
service
delivery
Social marketing
of ITNs
Condom social
marketing;
School youth
programmes for
HIV
Improving quality
of private drug
sellers;
School
deworming and
micro-nutrients;
Policies to reduce
indoor pollution,
information,
regulation;
Food fortification
laws with iodine,
iron, folate,
potentially zinc
Higher
tobacco
taxes;
Bans on
advertising
and
promotion;
Clear air
laws;
Counter-
advertising
Source: Jha and Mills (2002:52)
5.6 ‘THE WORLD HEALTH REPORT 2002’
This report focused entirely on interventions against the major risk factors, many of
which, at least amongst low mortality developing countries and developed countries,
relate to non-communicable disease (see Table 8). This improved on the earlier work
of Investing in Health which had acknowledged the limitations at that time with respect to
data and methodologies pertinent to risk factors.
30
Table 8: Leading 10 selected risk factors in developing countries, as
percentage causes of disease burden measured in DALYs
RISK FACTOR PERCENTAGE CAUSE OF DISEASE
BURDEN
High mortality countries
Underweight
14.9
Unsafe sex
10.2
Unsafe water, sanitation and hygiene
5.5
Indoor smoke from solid fuels 3.7
Zinc deficiency
3.2
Iron deficiency
3.1
Vitamin A deficiency
3.0
Blood pressure
2.5
Tobacco
2.0
Cholesterol
1.9
Low mortality countries
Alcohol 6.2
Blood pressure
5.0
Tobacco
4.0
Underweight
3.1
Overweight 2.7
Cholesterol
2.1
Indoor smoke from solid fuels 1.9
Low fruit and vegetable intake
1.9
Iron deficiency
1.8
Unsafe water, sanitation and hygiene
1.7
Note: Risk factors for which interventions are discussed in the report are highlighted in bold: further interventions are
due to be considered in a subsequent report
Source: World Health Organisation (2002:102)
The report highlights a range of interventions or clusters of interventions that are highly
cost-effective, but notes that their effectiveness varies according to circumstances (see
Table 9). Many of the interventions are built on behavioural change that requires
government intervention through, for example, legislation, tax or financial incentives. An
important finding of the report is that combining such interventions with more
individually-oriented ones is often highly cost-effective. The report does not propose a
‘package’ of risk-related interventions, nor discuss the relative priority of these
interventions vis-à-vis interventions proposed in earlier packages. However, the report
implies that a combined population- and individual-based approach is necessary to
prevent the advent of ill-health, especially in adults. Of concern with respect to non-
communicable diseases is that health systems in developing countries are seldom
geared to deliver interventions against these sorts of conditions, even at the level of the
individual. This will become increasingly problematic as the epidemiological transition
progresses in many countries.
31
Table 9: Conclusions on cost-effective interventions against risks
STRATEGY ASSESSMENT OF COST-EFFECTIVENESS
Strategies to protect the child’s
environment
• Cost-effective in all settings
• Very cost-effective components:
o Some form of micronutrient supplementation (depending on
the prevalence, vitamin A, iron or zinc)
o Disinfection of water at point of use to reduce diarrhoeal
diseases
o Treatment of diarrhoea and pneumonia
Preventive interventions to
reduce incidence of HIV
infections, including measures
to encourage safer injection
practices
• Very cost-effective, although care needs to be taken when
extrapolating the effectiveness of behaviour change from one setting
to another
• Use of some types of antiretroviral therapy in conjunction with
preventive activities is cost-effective in most settings
• (Directly observed therapy combined with testing for resistance not
cost-effective in all settings although there might be other reasons
for pursuing this strategy)
Improved water supply based
on disinfection at point of use
• Cost-effective in regions of high child mortality
Interventions to reduce the risks
of CVS
• At least one type of intervention cost-effective in all settings
• Population-wide salt and cholesterol lowering strategies are always
very cost-effective, singly and combined
• Combining them with an individual risk reduction strategy is also
cost-effective, particularly with interventions to reduce risk based on
assessed levels of absolute risk.
• The cost-effectiveness of the absolute risk approach could improve
further if it is possible to assess accurately individual risks without
the need for lab tests, and further work towards testing is
recommended
• (Increased physical activity was not evaluated but should be
considered as an additional strategy)
Source: World Health Organisation (2002:139-140)
6. THE CRITIQUE OF THE CONCEPT OF COST-EFFECTIVE PACKAGES
The concept of a cost-effective package, particularly as calculated and presented by
Investing in Health, has come under criticism from several quarters. These criticisms,
which fall into a number of categories, have particular implications for the design of
primary care services, forming as they do such a large part of basic packages. All the
criticisms raise the question of whether the recent dominance of cost-effectiveness as a
criterion for determining service delivery priorities is appropriate.
6.1 THE METHODS FOR CALCULATING THE BURDEN OF DISEASE
Williams (1999) and Mooney and Wiseman (2000) question the value of global burden of
disease estimates, largely because of the difficulty in attributing improvements in the
burden to changes in the health system. In terms of methodology, Williams (1999)
questions the value of cross-country comparisons given enormous differences in the
circumstances of different countries, and challenges the discounting, age weights, and
life expectancy approaches of Investing in Health. He questions, too, the values placed
on different states of ill-health, and particularly the role of experts in determining these
32
values, emphasising the importance of lay opinion in understanding the
community’s assessment of the impact of ill-health. Mooney and Wiseman (2000) raise
the need to reflect on values and how they are incorporated in such studies, appreciate
the contextual impact on values of the setting in which ill-health is experienced,
acknowledge the fact that changes in health are not valued equally by different
communities, and understand that an individual’s own assessment is not immune from
the norms of wider society. Thus, while burden of disease methodologies appear highly
technical, these commentators imply that they are highly ‘value-laden’. Consequently,
they fear that using burden of disease estimates to set priorities is both a diversion of
scarce research resources and likely to lead to inefficient and inequitable resource use.
They believe, like Williams (1999) that the real effort should be on determining the cost-
effectiveness of interventions, most particularly those related to new technology.
Murray and Lopez (2000) and Musgrove (2000) willingly admit to the limitations in the
Investing in Health methodology with respect to both the burden of disease and cost-
effectiveness calculations. They identify further research needs to include more data of
greater comparability, validity and reliability on the descriptive epidemiology of
conditions, as well as improved methods for, and wider consensus on, the estimation of
disability and application age weights. The greater use of sensitivity analyses is also
proposed. The World Health Organisation’s World Health Survey of 2002 is an example
of experimentation with new methodologies around self-assessment of health.
Nonetheless, these developments do not entirely address the concerns of the critics.
6.2 THE METHODS FOR ESTIMATING COST-EFFECTIVENESS
Investing in Health drew on the findings of a number of cost-effectiveness studies, but
did not scrutinise the quality of these studies in detail. Subsequently, the prohibitive cost
of these sorts of studies has led to a continuing interest in pooling data from different
studies in order to generalise cost-effectiveness conclusions. However, Walker and
Fox-Rushby (2000) found that only a relatively few studies on communicable diseases
(107) were published in the developing world between 1984 and 1997. Categorising
these studies they found that certain diseases and geographical areas had been
neglected and that appropriate analytic techniques were inconsistently applied.
The problems with analytic techniques were several. First, the studies generally
considered the health care provider perspective when calculating costs. This meant that
costs to other groupings, especially patients, were neglected. If these costs had been
included the conclusion of some of the studies may well have been different. For
example, Khan et al. (2002), using a social perspective for an economic study
conducted alongside a clinical trial at three sites in Pakistan, found that direct
observation of TB treatment by health-centre based health workers was the least cost-
effective strategy tested, because of the high costs of attendance faced by patients and
their escorts. DOTS raised patient costs by 267 percent for urban patients and 207
percent for rural patients. These high costs, as well as related factors,
12
seemed to be
12
Barriers to treatment as identified by patients included health-related problems, the time and
costs associated with a round trip, excessive waiting time at the treatment centre, the
unavailability of person to accompany them, social events (such as births, deaths, marriages),
33
deterring patients from attending. This was particularly so for patients who had the
most to lose from spending time under direct observation. Thus, one of the alternative
approaches, self-administration, emerged as most effective. A sub-group that was
observed directly by community health workers achieved the highest cure rate whilst
being only slightly more expensive than self-administration. Khan et al. (2002:178)
conclude that, ‘without stronger evidence of benefits, it is hard to justify the costs to
health services and patients that this type of direct observation imposes.’ This case
study illustrates the importance of perspective in economic evaluation, and also makes
the point that many health service activities have become standard practice without
evaluation.
The other analytic problems identified by Walker and Fox-Rushby (2000) are listed in
Box 6, and apparently reflect similar problems identified by reviews of studies performed
in the developed world. A review by Mulligan, Walker and Fox-Rushby (2003) of 32
economic evaluations of non-communicable disease interventions in developing
countries (which had been published between 1984 and 2000) reinforces the findings of
Walker and Fox-Rushby (2000). In addition, it was found that very little attention was
paid by studies to the issues of affordability and sustainability, and that studies tend not
to support policy decisions especially as outcomes are not measured in a way that can
allow the comparison of range of interventions across health sector.
Box 6: Analytic problems with economic evaluations of communicable and
non-communicable disease interventions (1984-2000)
Source: Walker and Fox-Rushby (2000), Mulligan, Walker and Fox-Rushby (2003)
6.3 THE USE OF COST-EFFECTIVENESS FINDINGS TO DETERMINE SERVICE
PRIORITIES
6.3.1 The generalisability of cost-effectiveness findings
Paalman et al. (1998) question the usefulness of global exercises to inform national
priority-setting, because of the influence of local economic, political, cultural and
infrastructural factors on health systems, access to health care, and the burden of
job/occupational reasons, the unfriendly attitude of staff, and the lack of support by ‘significant
people’.
1. The studies generally considered the health care provider perspective when calculating
costs. This meant that costs to other groupings, especially patients, were neglected.
2. There was a reliance on intermediate outcome measures, which means that the cost-
effectiveness with respect to achieving health status improvements had not been
calculated.
3. Bias was evident in the sorts of costs that had been included. Donated costs, and some
capital costs, had often not been taken into account.
4. The impact of time – both in terms of the annualisation of capital costs and the use of a
discount rate – had been inconsistently applied.
5. There was a lack of transparency with respect to the sources of data.
6. There was an absence of critical examination of the findings, with many papers failing to
perform a sensitivity analysis.
34
disease. Mulligan, Walker and Fox-Rushby (2003:1,12) conclude from their review
that ‘the quality of studies was poor and resource allocation decisions made by local and
global policy-makers on the basis of this evidence could be misleading … Overall, it
seems that the existing evidence base from developing countries is very unlikely to be
able to challenge effectively any resource allocation decisions made on the basis of
burden of disease estimates without a great deal more investment in the number of
economic evaluations across countries and health interventions.’ The ‘WHO-CHOICE’
project is engaged in developing improved standards for cost-effectiveness analysis as
well as league tables that allow comparison of results between regions, but Mulligan,
Walker and Fox-Rushby (2003) make the point that research studies should not just
increase in number. More research needs to be performed that investigates the causes
of variation among cost, effect and cost-effectiveness data within and between settings
(some of these are presented in Box 7, many relating closely to the cultural and
economic contexts of settings). More research also needs to be performed on
assessing and testing the transferability of findings between settings. A recent WHO
randomized trial of a new model of antenatal care is cited by Fox-Rushby (2003) as an
example in this regard. Until the evidence base is much more secure, then, the
implication is that cost-effectiveness information should be used with circumspection in
priority setting. As mentioned before, it is probably most useful in assessing new
technology, and in refining existing interventions.
Box 7: Causes of variation in cost-effectiveness calculations between
settings
Source: Adapted by Mulligan, Walker and Fox-Rushby (2003) from Jamison et al. (1993)
6.3.2 Primary care as integrated and responsive
For several critics of Investing in Health, cost-effectiveness simply cannot capture the
value of primary care services as envisaged by the ‘Primary Health Care Approach.’
First, Paalman et al. (1998) feel that, despite the concept of ‘a package,’ the listing of
individual interventions will promote vertical and ineffective programmes. For Pinotti et
al. (2001), this leads to missed opportunities for screening, treatment and diagnosis for
conditions that are not necessarily the presenting complaint of the patient. In turn, this
prevents the achievement of economies of scale and scope, and wastes opportunities to
prevent further ill-health at an early stage. However, for Bobadilla et al. (1994)
Epidemiological environment
Prevalence of condition
Incidence of condition
Existence of competing risks or synergisms
Individual characteristics
Age
Tendency to compliance
Tendency to self-refer
Level of risk factors
Individual variation in values
System characteristics
Local cost of non-traded inputs to health care system
Generalisable systemic competence
Discount rate
35
verticalisation was not the intention behind the package, and Musgrove (2000) feels
that the impact could be quite the opposite (see the examples of Bangladesh and
Tanzania later). Musgrove (2000:113) argues that ‘the reason for including the size of
the burden as a criterion for priority setting is precisely that health systems in poor
countries often cannot efficiently administer a large collection of programmes, and
dissipate their resources trying to do so. In consequence, it makes sense to maximise
the health gains from a small number of interventions, economising on scarce
managerial and administrative capacity.’
However, the tendency of governments and donors to promote vertical programmes
must be acknowledged. As Pinotti et al. (2001:76) provide by way of an example, ‘we
cannot ignore the many obstacles to initiating and achieving an integrated approach to
reproductive health care. The medical community has not always reacted positively to
the delegation of functions. The structure of the public health system is not well adapted
to this kind of approach either, but tends to prefer to deal with one health problem at a
time …International organisations are always pressing developing countries to deal with
women’s health in terms either of family planning projects of safe motherhood projects,
while considering a more integrated approach as too complicated and ambitious and
objective.’
A second argument against cost-effectiveness as a predominant criterion in priority-
setting is one that contests that health services are purely concerned with the reduction
of death and disability. Critics ask what is to made of other criteria, such as ‘the rule of
rescue,’ respect for patient autonomy, a dignified death, the satisfaction of having
access to screening services, and wellbeing improvements arising from the knowledge
that services are universally available. Mooney and Wiseman (2000:370) write that
“there is so far to go to embrace all that is relevant. Care is needed to avoid driving out
these intangible issues, and letting the measurable dominate or even monopolize the
objective function … There is no reason why a benefit or cost that is difficult to measure
should be considered less important than one that is easily quantifiable, and to give
priority to generating easily measured benefits is to distort priorities systematically.’ The
key question for Mooney and Wiseman (2000:370) is understanding ‘what is the nature
of the good that the community wants delivered by its health services.’
Paalman et al. (1998:24) concur that ‘the package arrived at by experts might not be
acceptable to the public, and hence less cost-effective in the end.’ Pinotti et al. (2001:
70) provide the example that, ‘although the main complaints women had [at a hospital
outpatients department in Brazil] were vaginal discharge, menstrual or sexual disorders,
need for contraception, abdominal pain and breast lumps, national priorities were linked
with maternal mortality and family planning.’ Musgrove (2000:114) agrees that ‘this
raises the general, and severe, problem of needs and wants not necessarily coinciding,
and points to the urgency of finding out whether and why people may not use existing
health services and reforming those services so at to attract the beneficiaries they are
meant to help.’ Although this point may go some way to assuage critics’ concerns, it
does not fully acknowledge that some services at the primary level may have high value
despite the fact that they are not highly cost-effective.
36
6.4 ACHIEVING EQUITY
Paalman et al. (1998) note that ‘the fact that the most efficient interventions … tend to
specifically benefit the poor is more a result of coincidence than of principle.’ Musgrove
(2000) quotes Gwatkin and Guillot (1998) as substantiating this finding more fully. Thus,
the cost-effectiveness approach does not intrinsically protect equity.. Indeed, as
mentioned earlier, cost-effectiveness principles might argue against the extension of
services to populations in remote areas, as the cost of delivery increases in such
circumstances. Investing in Health (World Bank 1993), Bobadilla et al. (1994) and
(Musgrove 2000) all emphasize that governments need to make explicit choices
between equity and efficiency concerns. This trade-off is easier to manage in wealthier
countries where resources are not so scarce. In addition, when the cost-effectiveness
approach is combined with the burden-of-disease approach, as advocated by Investing
in Health, the equity-efficiency trade-off is tempered. These authors also argue that,
access by the poor to cost-effective services can be promoted through a variety of
financing mechanisms that give priority to the poor either because they are sicker or less
able to pay. While they recognise the importance of incorporating distributional
concerns within priority-setting processes, they emphasise that within poor populations
services should be prioritised only according to cost-effectiveness considerations.
Clearly, then, the Investing in Health approach is based on the principle that health care
interventions, rather than population groups, should be prioritised. This ignores the
other dimensions of priority-setting, especially equity considerations.
This approach concerns several critics, for whom the principles of vertical and horizontal
equity should receive high priority (for example, Paalman et al. 1998, Almedia et al.
2001). The suspicion that the cost-effectiveness approach might run counter to equity
goals is compounded by the association the World Bank (and particularly Better Health
in Africa) has with the concept of user fees, a financing mechanism which in some cases
had a negative impact on equity (see, for example, Gilson 1998).
Rannan-Eliya (2001:37) rejects the cost-effectiveness approach, stating that ‘cost-
effectiveness of interventions and a disease-focused approach to allocational efficiency
are irrational and inefficient guides to resource allocation and may lead to erroneous use
of resources.’ He states on the basis of the Sri Lankan experience that ‘unless equity of
access is the highest priority, choices about rationing will be made which inevitably hurt
the poor’ (Rannan-Eliya 2001:36). Sri Lanka enjoys a superior health status for its level
of economic development. This has been achieved without reliance on the cost-
effectiveness approach: from 1930-1985, ‘despite spending less than 25 percent of its
health budget on [minimum cost-effective package] services and less than one-eighth in
dollar terms of the minimum expenditure level [set by the World Bank], Sri Lanka
achieved more than double the DALY-gain promised by the World Bank/WHO’
(Rannan-Eliya 2001:30). Rannan-Eliya (2001) ascribes this success to a mixture of pro-
equity policies, the tailoring of services to take account of what is available in the private
sector, and the provision of hospital care to mitigate the impact of catastrophic illness.
This last point – the need to mitigate the impact of catastrophic illness - takes account of
the welfare, and not simply health, benefits of health care interventions. For example, a
review of cost of illness and coping strategies (with a focus on malaria, tuberculosis and
HIV/AIDS) (Russell 2003) found that the direct costs of illness to households was
37
generally between 2.5 and 7.0 percent of household income, while the indirect costs
of illness were often as significant – or even more significant – than direct costs. Ensor
et al. (2002:255) comment that “it is therefore apparent that while more resources for
ESP [essential service packages] may increase the use of effective primary services, it
has very little impact on household exposure to financial risk through illness.’
However, whereas there is a strong argument for prompt and appropriate hospital-based
care to allow some of these costs to be avoided, in the long-term one would want
primary care and other services to reduce the need for hospital care. For example, the
World Bank (1994:5) found that ‘when offered by well-functioning health centers, the …
package has reduced total hospital admissions in some communities in Africa by up to
50 percent and has cut hospital admissions for such illnesses as measles, tetanus and
diarrhoea by up to 80 percent’.
6.5 IMPLEMENTING PRIMARY CARE SERVICES
While critics of the cost-effectiveness approach – or at least of its dominance – worry
about its value basis and methodological flaws, implementers face the vexing question
of how to deliver services on the ground. As late as 2001 Pappas and Moss (2001) write
that ‘despite many efforts at the WHO and other agencies, primary care services have
been inadequately funded and poorly maintained in many countries, and the results
have been disappointing. Sustainability of health systems has often been impeded by a
lack of health policy and management.’
The challenges to implementation appear so vast (see the case study from Cameroon
presented in Box 8) that the finer details of cost-effectiveness controversies must seem
almost immaterial to managers working at the district level. Simply improving the
effectiveness of district health systems would automatically improve the cost-
effectiveness of interventions in general. Some of the frustration felt by critics of the
cost-effectiveness approach appears to result from a sense that research efforts are
disproportionately focused on perfecting burden of disease estimates and economic
evaluation techniques, when the more pressing problems lie within the field of health
systems research. In commenting on the experience of the IMCI strategy, Bryce et al.
(2003) draw the important distinction between individual interventions and ‘delivery
strategies,’ commenting that more intensive efforts need to be focused on improving the
latter.
The next section examines some country experiences of the implementation of
packages, in order to assess the extent to which health system problems impact on the
usefulness of the concept. Suffice it to say here that many factors need to conspire to
ensure successful primary care delivery. Thus, for example, Nitayarumphong (1990)
attributes the success of primary health care in Thailand to an array of factors, including
community involvement in health, collaboration between government and non-
government organizations, the integration of the PHC programme, the decentralization
of planning and management, inter-sectoral collaboration at operational levels, resource
allocation in favour of PHC, the management and continuous supervision of the PHC
programme from the national down to the district level, as well the ‘horizontal training’ of
villagers to villagers.
38
Box 8: Problems in implementing primary care services: the case of
Cameroon
Source: Adapted from Essomba, Bryant and Bodart (1993)
7. THE IMPLEMENTATION OF MINIMUM PACKAGES: THREE CASE
STUDIES
Ensor et al. (2002:248) note that ‘it is a little surprising that so little evaluation [of
essential service packages (ESPs)] exists, given that the introduction of ESPs might be
seen as the largest experiment in the use of evidence-based clinical practice and cost-
effectiveness analysis in priority setting within the health sector.’ Indeed, this review
could only find two published evaluations – one by Ensor et al. (2002) in Bangladesh,
and the other by the Bamako Initiative in Benin and Guinea.
13
A third evaluation would
13
A report does exist evaluating the overall impact of the Bamako Initiative but this review was
unable to access it.
Limitations of the PHC policy adopted in 1982
following the Alma Ata Declaration
• Problems within the health services (e.g. poor referral systems, neglect of health
promotion activities at the expense of curative services, inconsistent care due to poor
training)
• Problems at the community level (e.g. lack of support for the concept of community health
workers, pressures to upgrade facilities, sustainability problems resulting from community
health workers requiring reimbursement)
• Problems with the interface between the community and the health services (e.g.
failure of the community health committees, lack of consultation with community health
workers)
• Problems at the delivery level in the health sector (e.g. poor co-ordination, supervision
and training)
Limitations of the re-orientation of primary health care in 1989
• Inadequate legal framework (e.g. lack of legislation regarding cost recovery based on the
sale of medicines, the formal organisation of communities into committees and revenue
retention, as well as the absence of a formal national health policy)
• Incompatibilities between the political structure and the new health structure (e.g. the
different definitions of districts)
• Incompatibilities between the goals of the new health policy and the organisational
chart of the MOH (e.g. inadequate restructuring, overlapping functions)
• Absence of a critical mass of trained personnel in health management
• Too highly centralised management with poor co-ordination of human resource
management
• The slowness of the extension of the primary health care coverage
• Inability of the system to assure the availability of, and easy accessibility to,
medicines
• Inadequate health information system
• Poor promotion of the new policy
• Poor co-ordination of operations research
39
be that of ‘TEHIP’ (the Tanzania Essential Health Interventions Project), which
utilises a number of decision-making tools in planning and refining district services, but
this has not yet published final findings. However, this review is able to present a brief
overview of the impact of TEHIP written by Don de Savigny, former Research Manager
for TEHIP.
7.1 THE BAMAKO INITIATIVE:
The Bamako Initiative was launched by the African Ministries of Health. It built on
regional experience following Alma-Ata which raised the issues relating to the delivery,
affordability and sustainability of primary care services. The initiative, which was
adopted in several countries, combined the concept of cost-effective minimum care
packages with health centre and district revitalisation, rationalization of resource use and
management, and community outreach, co-management and cost sharing. In Benin and
Guinea the effectiveness of the programme – and of the integration of curative and
preventive services - was demonstrated by the simultaneous increase of coverage for all
the interventions covered in the package. Importantly, this was achieved within existing
resource constraints. Further, the gradual improvement of coverage between 1989 and
1993 showed that reorganized health centres can yield benefit for many years.
Box 9 provides an extensive description of how this was achieved. A central
characteristic of the approach was the development and implementation of the package
taking account of local circumstances. New interventions were included over time in
response to both public health needs as well as community demands. Levy-Bruhl et al.
(1997:77) state that ‘the dynamic process of local and participatory problem-solving
approaches through monitoring of coverage, identification of problems and micro-
planning of solutions is probably one of the most important factors for success.’
The median cost per capita per year of delivering the package was US$1.2 in Benin and
between $0.7 and $1.0 in Guinea which had lower salaries and incentives (2001 prices).
These costs were consistent with the experience of the Bamako Initiative in other
countries (see Table 10 (Table 8 in Appendix 2 gives the costs in the original 1993
prices)). In Benin and Guinea, the cost varied very little between regions, cost centres
and over time. Variations in cost-effectiveness were caused by the levels of coverage
achieved, and improved over time, as did affordability. Soucat (1997) concludes that
‘the evidence shows that integrated PHC of acceptable quality can be delivered by
health centres for around $1.2 to $2.5 per person per year’, and notes that this is far
lower than the Better Health in Africa estimate.
14
Soucat (1997) attributes this difference
to the integration of services in the Bamako Initiative sites (the World Bank had compiled
its estimates by summing the costs of national programmes for different interventions),
and predicted that immediate cost-effectiveness gains would be felt if new services were
added to existing services in Benin and Guinea, because of economies of scope.
14
It is difficult to know to which aspects of the Better Health in Africa estimates these figures are
being compared. The direct health centre costs estimated by Better Health in Africa were $6.2
per capita, but this excluded support by the district management team, which was included in the
Bamako Initiative estimates. However, the Better Health in Africa costs may have included more
extensive services at the health centre level. The costs estimated by the Bamako Initiative
appear to represent the full costs faced by health services in delivering services (and not the
costs after cost-recovery from patients, a strategy for which the Bamako Initiative is best known).
40
Box 9: Operational strategies to improve effectiveness under the Bamako
Initiative in Benin and Guinea
Source: Knippenberg et al. (1997b), Levy-Bruhl et al. (1997)
Table 10: Health centre operating costs per capita and per year within the
Bamako Initiative in some African countries, 1989-1999 (2001 prices)
COUNTRY TOTAL COST PER CAPITA ($)
Benin 1.20
Guinea 0.82
Guinea-Bissau 0.70
Mali 1.51
Senegal 2.91
Zaire 1.29
Cameroon 0.33
Source: Soucat et al. (1997)
(i) The essential health care package was developed.
(ii) Health centres were identified as the basis for revitalizing the health system as they are the first
point of contact.
(iii) Obstacles to availability, accessibility and utilization of health services were identified by health
centre personnel, community groups, the Ministry of Health and donors.
(iv) Activities were prioritised to respond to the most important causes of morbidity and mortality in
each area in order to maximize the impact of health status with very limited resources (mainly
focusing on the most vulnerable members of the population, pregnant women and children):
o childhood vaccinations;
o growth monitoring of young children;
o antenatal and prenatal care (including tetanus toxoid vaccination, iron supplementation,
and routine malaria prophylaxis);
o oral rehydration for diarrhoea;
o curative care: malaria and acute respiratory infection and, in response to community
demands, other curative services were added such as treatment of parasites and sexually
transmitted infections;
o Vitamin A deficiency reduction was not included initially because related health problems
were minimal, although Vitamin A supplementation was introduced later; and
o STI and HIV/AIDS treatment was later included in Benin in response to community
demands.
o (Promotion of breastfeeding was not included because related health problems were
minimal).
(v) Curative and preventive care responsibilities were integrated so that each health worker dealt
with all aspects of care provided to each patient.
(vi) Outreach sessions had a positive impact on effectiveness. However, outreach became difficult
where populations were very scattered, which prevented sufficient frequency of visits to offer
continuity. This demonstrated the need for intensified follow-up and, in some cases, additional
infrastructure and personnel.
(vii) The continuity of care was improved through improved tracking of patients.
(viii) The availability of essential resources such as drugs and vaccines was improved through
achieving good local management and proper functioning of logistical support, built on the
consensus of partners at all levels to support the initiative.
(ix) Diagnostic and treatment practices were improved through the development of decision trees
with active involvement of local nurses, as well as the development of risk screening.
(x) Referral infrastructures to district hospital were organised for obstetric emergencies.
(xi) Monthly supervision on all aspects of health centre functioning was organised.
41
This apparent success was not without its problems, however. Equity problems were
experienced with respect to access to curative services, and the location of some health
centres distant to the populations they serviced posed a constraint, outreach services
notwithstanding (Knippenberg et al. 1997a). In addition, the fact that curative utilization
rates stagnated at 0.3 visits per capita per year raised the question of whether the
package was responding adequately to community needs. Knippenberg et al. (1997a)
identified the need for more research into this area, especially in relation to diarrhoea
and malaria, as well as the need for further quality of care improvements.
7.2 BANGLADESH
The Bangladesh experience is not as unequivocal as that described by the Bamako
Initiative. In 1998 Bangladesh began a sector-wide approach to extending health care to
vulnerable populations, especially through an essential service package emphasising
maternal care, certain communicable diseases and child health (see Box 10). The
package was designed to improve population health status through a targeting approach
which singled out facilities used more by the poor, effective services for diseases borne
proportionately more by the poor, and rural areas where population health is the lowest.
That is, the assumption was that improvements in health status could be supply-led.
Evidence for the components of the package was mainly drawn from international cost-
effectiveness studies, although some studies were conducted locally, whil