World Health Organization logo World Health Organization
Author: Enrico Pavignani

Analysing disrupted health sectors: a toolkit
Module 12: Formulating strategies for the recovery of a disrupted health sector


The module discusses practical ways to approach the recovery of a disrupted health sector, suggesting step-by-step iterations aimed at appraising and costing merits and drawbacks of different broad options available to policy makers. Ways to project the effects of conservative recovery strategies, alongside those induced by the adoption of alternative service delivery models, are described. Flaws commonly found in troubled health sectors and possible policy responses are sketched. The methods described are supposed to be applied to a transitional context, such as during the last years of a conflict, when peace negotiations are under way and a final settlement is anticipated. This module assumes familiarity by the reader with most of the issues covered by Modules 2 to 11. For the sake of brevity, issues and methods discussed in detail in other parts of the toolkit are only briefly mentioned.

Annex 12 presents summaries of some already completed reconstruction processes, proposed as empirical reference frames to decision-makers involved in transitional health sectors.

Recovery after extensive destruction may offer a unique chance to reconsider the whole health sector and plan it on a comprehensive, rational basis. In many instances, large amounts of capital become frequently available to address major allocative distortions; the atmosphere of change may reduce resistance aimed at preserving the status quo; massive destruction and dilapidation make the abandonment of unwanted facilities easier. Thus, building an equitable and sustainable (in the long-term) health system may become a realistic target. A country emerging from a prolonged crisis cannot afford to miss that chance.

Furthermore, in the optimistic mood that usually characterises reconstruction, investment decisions should not be taken light-heartedly. They will shape the health sector far into the future. In the same way present allocative decisions are heavily influenced by investment choices made decades ago, the future allocation of recurrent resources will follow to a large extent the distribution of the physical infrastructure resulting from the recovery.

The rethinking of the health sector should go beyond the hardware that usually absorbs the attention of decision-makers. To take advantage of the opportunity and to complete the reassessment of the sector, legal, regulatory and management systems, as well as health care delivery models, need equivalent revisiting. If the rethinking of the sector is based on a robust understanding of the situation, success can follow. Conversely, policy changes inspired by imported and often untested recipes imposed by outsiders, whose only merit is the power they hold, are frequently unsuccessful. A sensible and effective recovery strategy is likely to emerge from the balanced encounter of insiders knowledgeable of the country and outsiders familiar with potentials and pitfalls of transitional processes.

Many health sectors grow organically over the years, being shaped by countless political and economic decisions unrelated to each other. Even in countries where PHC has dominated the policy discourse over a long time, certain distortions, such as the hospital bias, may persist unabated, because of strong interests. Additionally, protracted crises tend to hide existing distortions, which when left unrecognized can only get worse.

In most instances, the recovery of a disrupted health sector calls for:

Thus, to revamp a crippled sector means to increase its equity, effectiveness, appropriateness and efficiency. These dimensions can receive different degrees of attention, according to the perceptions and priorities of decision-makers and the constraints conditioning their actions. For instance, to expand service provision to cover previously destitute populations may appear more attractive to politicians than to health professionals, who are often more worried with the quality of the offered care. Local grievances may be appeased with health infrastructures, financed by aid agencies, as a peace-building measure. Ruling elites, mainly concerned with their own welfare and their own constituency, are prone to give precedence to tertiary care in the capital town. MoH officials may perceive the recovery process as an opportunity to claw power back from aid agencies. Other stakeholders, such as external financiers, may emphasize efficiency considerations, or, when under pressure to demonstrate progress, just give precedence to actions that are visible, easy and fast to implement. Frequently, most parties support physical rehabilitation, while other aspects of health care delivery are neglected.

Learning from previous recovery processes

Despite the efforts of researchers and practitioners over the last two decades, learning from previous conflict and post-conflict experiences has not thriven. Responses to new crises range from the mechanical replication of previous approaches, to the starting anew every time, in this way discovering again some well-known lessons. The difficulty of comparing different situations and of drawing correct, appropriate lessons from them is not exclusive of conflict-affected health sectors.

Investigating the field of health sector reform, McPake and Mills (2000) argue that the propensity for either none or too much transfer of approaches reflects two symmetric fallacies: “the search for a single best model” on the one hand, against “the belief that nothing can be learned from other contexts” on the other hand. They suggest a way out of this unhelpful dichotomy, by using a conceptual framework, which recognises three groups of valid conclusions: a) generalizable to most or all situations; b) specific to a given context; and c) valid for a sub-group of situations, considered sufficiently similar. Clarifying the nature of the conclusions should prevent decision-makers from transferring a b-type conclusion to another crisis, or from transferring a c-type conclusion to a crisis of different character. The snag of this sensible approach is that it demands to decisions-makers a robust knowledge of both previous crises and of new ones, a knowledge that very few involved players hold.

Objective constraints limit the access to the required knowledge. A first group of problems lies with the nature of the lessons learned. Learning from crises is difficult, even for insiders, who are exposed to fragments of evidence, rarely assembled into a coherent picture. Thus, comprehensive lessons are rarely learned, whereas incomplete ones may be wrong or misleading. Further, the shortage of solid, widely accepted information makes most lessons arguable (particularly the controversial ones). Parties displeased by a certain conclusion find always room for demolishing it. Politics and vested interests inform and sometimes distort knowledge, hence limiting its transferability.

A second group of constraints relates to the way knowledge is managed and exchanged. Available knowledge is dispersed across agencies, research centres and NGOs. Much knowledge, held by people, remains unwritten. And written knowledge tends to get lost in the crisis environment. Language barriers, frontlines, mistrust, displacements, short assignments, all hinder the exchange of information among participants.

A third group of difficulties is linked to actors. The insiders of a health sector entering a crisis are usually not conversant with the issues, nor with the international debate related to health care provision in conflict and post-conflict settings. Conversely, outsiders, failing to understand the specific features of a new crisis, bring with them the lessons learned elsewhere, which may not apply to the new context. Researchers are in a better position to facilitate the transfer of knowledge, but they convey it according to academic etiquette, hardly the most appropriate communication tool, given the target audience. Additionally, they may publish their findings with years of delay, when decisions have been already made, and tend to remain in country only during short time spans.

A fourth group of obstacles relates to the way the aid enterprise is structured. Donor agencies have been singled out as poor learners (Berg, 2000). They shape the policy debate, often undermining learning from previous crises, by imposing corporate policies. These lack sometimes a strong empirical basis, may be inappropriate to the local context (Strong, 2003), and are usually presented stripped down of assumptions and caveats to policy-makers. Also, by fragmenting the health field and flooding it with their own priorities, donors raise the level of noise to such an extent that learning, both from inside experience and from outside knowledge, becomes impossible. Short programming cycles and assignments compound the picture and distort learning: what looked promising at a given point in time might emerge as a flop later on, when some unanticipated side effects emerge. In this way, wrong lessons may be drawn or right lessons be neglected.

The host of barriers to learning from experience is discouraging. Some of them, of a structural nature, cannot be fully overcome, but just controlled to lessen their impact. Other constraints might be addressed only by a radical change of the way the aid system operates. Hence, the most likely outcome of future crises is the familiar constellation of wrong lessons applied, right lessons ignored, insensitive decisions made and ineffective action pursued.

‘Best practice’ is unlikely to be found embodied in a specific ‘best’ policy or model. Rather, it emerges from a judicious balance of context-sensitive exploration, rational appraisal of alternatives, restrained generalisation of specific experiences. The same approach, to be applied across different crises, should lead to the drawing of different conclusions and to generating different policies. The lesson lies in the methods adopted to pursue an adequate understanding of the picture, rather than in the choices triggered by such an understanding.

For an excellent discussion of learning in the humanitarian field, see Van Brabant (1997). Annex 12 offers to the consideration of interested parties some condensed reviews of documented recovery processes. An insightful, detailed discussion of these issues may be found in Bower (2002). By looking at the variety of situations and responses, the actors of future recovery processes should be able to recognise similarities as well as differences, and make decisions accordingly. For a discussion of the country context and its relationships with the health sector, see Module 3.


At the onset of the strategy formulation process, precision is never possible and to some extent is usually unnecessary. Most countrywide allocative decisions are by their nature aggregate and approximate, thus robust in relation to the imprecision of the estimates upon which they are based. For example, preliminary figures may suggest that a neglected area need a dramatic overhaul of its PHC network. Rough calculations estimate the existing gap in the order of US$ 4-6 million. The decision to be made is whether to encourage an NGO, able to invest in PHC-facilities about US$ 600,000 per year during 3-5 years. Whatever is the gap size computed from accurate figures eventually obtained, it will be substantially reduced by the NGO intervention. The essential feature of the initial analysis is its accuracy in relation to main problems and constraints. In other words, decision makers need to be reasonably confident that a given major problem, such as the inadequacy of the PHC network in the example mentioned above, is not an artefact, bound to disappear once data precision improves. The exact quantification of such a problem can wait for a later phase, when corrective measures are actually introduced and better data are gathered.

At the first attempt of going through the steps suggested below, it will become painfully patent that many of the required figures are not available or are seriously flawed. The manipulation of the available data will contribute to detect their shortcomings and will provide a powerful stimulus to strengthen them. By commissioning dedicated studies, when this is feasible, and revisiting source data, so as to strengthen available estimates, the information base must be strengthened. Periods of lull in the crisis, as during peace negotiations, may offer a precious opportunity for building an enhanced information basis and to start recovery with mature, agreed-upon plans. While the needed studies proceed, educated guesses can be used, provided their inadequacy is recorded and future users of the projections obtained in this way are made aware of the caution demanded when using them. As soon as better data become available, the projections must be revisited.

To complete sound recovery plans, a timeframe of 1-2 years can be anticipated, depending on the baseline situation and the complexity of the health sector. In some cases, sudden political or military developments hasten the pace of the recovery strategy formulation process, which must take place within months rather than years. Whereas the overall conceptual approach remains the same, a recovery strategy must emerge as soon as possible, to inform pressing decisions that cannot wait. Avoiding that catastrophic mistakes are made in the frantic climate of certain hurried transitions must be the main concern of those involved in them.

The initial round of exploration of the available information and its consolidation may take some months, during which informants and stakeholders are contacted and involved. Precious clues about prevailing perceptions and preferences are obtained. The main recovery directions may emerge at this stage. The findings of this exploratory round may be condensed in a health sector profile highlighting the main problems faced by participants (for details, see Module 13). Additionally, an interim recovery strategy, which suggests possible ways forward, makes their implications explicit and points to the main information gaps to be filled, may be sketched. Measures deemed urgent, or clarified to such an extent that dedicated studies are not mandatory, may be introduced already at this stage. Health sector profile and interim recovery strategy should be conceived as discussion and negotiation tools, and be written in a way accessible to most stakeholders.

The second round, consisting of studies considered as essential to put the policy discussion on firm grounds, may demand a longer period, say 6-12 months. The responsibility for carrying out these studies can be distributed among players, to share the burden and increase participation. At this stage, the temptation of studying most aspects in detail must be resisted. Given the fast pace of change typical of transitional contexts, most details will become outdated before they have been used to inform action. The needed studies must explore the field only to gather intelligence valuable for the decisions to be made in the short and mid-term. Detailed studies must be programmed for later stages, when the sector has stabilised and the planning horizon has expanded. Additionally, the studies must be judiciously spaced, to encourage actors to participate and help them to absorb findings.

In the third round, new inputs are consolidated in a set of alternative projections, to be submitted to decision makers. The ensuing open debate may lead to revised projections, as unforeseen aspects are considered or trade-offs are agreed. Once a measure of consensus and support is attained, strategies can be finalised and formally endorsed. Finally, they have to be translated in operational plans, which integrate the contributions of most participants into a consistent framework.

This approach calls for the establishing of permanent in-country capacity, so as to strengthen the previous work in light of its limitations and consistently with the adopted methods. Unfortunately, continuity of work is a rare, fortunate event. More often, projections are elaborated during short periods of intense activity by visiting consultants. The limitations and ‘heroic’ assumptions built into their results are quickly forgotten and their conclusions are taken uncritically, at face value. Alternatively, their work is superseded by new developments; other consultants are called in to elaborate new projections, which risk the same degree of oblivion met by the previous ones.

The main responsibility for the proposed work should obviously lie with the government, despite the weaknesses it may suffers. To embark in an exercise along the lines discussed in this module will attract competent cadres and encourage the emergence of some capacity. If successfully carried out, the exercise will boost self-confidence within the government and improve its standing with development partners. In very special situations, no ‘government’ is in place, or it is too contested, to play a useful role. Hence, interim authorities and aid agencies must assume the bulk of the responsibilities. Local participation should be pursued to the largest possible extent.

Scenarios can be built by following a top-down approach, by starting with a consideration of the global financing envelope and deducing from it what services will be affordable. This approach suits better severely disrupted contexts, where health care is fragmented, health authorities are absent or incipient, and most information is not available. In these conditions, approaching the analysis in aggregated terms may represent the only realistic option. Also, a top-down approach may be indicated in situations of urgency, when additional data cannot be collected. Examples of situations better studied through a top-down approach: Afghanistan in 2002 and Southern Sudan in 2003.

Alternatively, the analysis may start the other way round, following a bottom-up approach, considering the facility unit recurrent cost and progressing to compute the total expenditure of running the whole health sector. This more information-intensive approach looks appropriate to distressed, but not collapsed and fairly stable health sectors. The existing information base, although deficient, may provide a starting point for the analysis. In some cases, a stalled peace process may offer the opportunity to collect the missing data, in this way enhancing the results of the exercise. Example: Mozambique in 1990-92.

The computations proposed below proceed iteratively. Usually, several rounds are needed to reach acceptable results. The computations are presented in several sequential steps, to convey the logic of the process, but do not need to be carried out necessarily in the same order. Convenience and availability of data may suggest a different sequence. The eventual results should not differ significantly. The steps proposed are common to both approaches; they are discussed only once for the sake of brevity. When feasible, to approach the exercise from both sides is recommended, on learning and consistency grounds. In this way, the resulting final estimates will gain robustness.

Top-down approach

Step one. Estimate the present level of aggregate financing, total and per head. Include all sources of financing. Private contributions, often unknown, can be very significant and should not be neglected. They vary dramatically across countries, and, given the dispersion of provider-patient transactions, are difficult to estimate. Censuses or household surveys, which usually supply this information, are in most cases not available, or cover only secure areas, i.e., very unrepresentative ones. Again, the judicious consideration of countries considered similar in terms of socio-economic development (but not victims of a serious crisis) can provide some indications.

Step two. Try to forecast the level of internal and external financing to be allocated to the health sector in the mid- and long-term, given macroeconomic perspectives (usually studied with some accuracy, by IFIs, independent analysts, donor agencies etc.). This depends on several factors, including economic growth, the government capacity to extract revenues, the priority given to health by decision-makers, the popularity of the country within donor circles etc. Build a set of scenarios (high-case, average and low-case). See Module 6 for a detailed discussion on forecasting financing levels.

Consider that in most war-torn countries fiscal capacity has suffered badly, when it has not altogether collapsed. Recovery from fiscal collapse is usually slow. Also, the substantial cost of the peace process, likely to be felt far in the future, may offset anticipated peace dividends. And other important sectors compete with health for government attention. Thus, be wary of over-optimistic forecasts, quite common in transitional environments, when expectations are high and the implications of rebuilding a devastated country are not appreciated in full.

Step three. Compare the global resource envelope likely to be available to the health sector in the mid- and long-term to equivalent figures for other countries and analyse what they have achieved. There is no reported example of a very poor country able to provide universal coverage with comprehensive basic services of acceptable technical quality to its citizens. A realistic estimate (Hay, 2003), which puts the annual minimum cost of a comprehensive, universal publicly-financed health care at between I$ 75 and 120 per capita, goes a long way towards explaining why the goal of universal coverage is out of reach for the poorest countries. Thus, the figure arrived at for the total resource envelope, which in most cases of poor, war-torn countries falls between the level of Afghanistan (2-3 US$) and that of Cambodia (22 US$), imposes on policy-makers very harsh decisions, in terms of scope, coverage, content, quality of the provided services.

Step four. Study the composition of health expenditure and assess whether it is balanced (in most cases it is not). In many disrupted health sectors, the information related to health expenditure is grossly inadequate, and only educated guesses are allowed. During the first years of physical reconstruction, investment expenditure may expand to absorb up to one third of the total, but later it should stabilize at below 20%. In a labour-intensive sector such as health, salaries should account for between half and two-thirds of recurrent expenditure. The balance of recurrent expenditure, after salaries are subtracted, may be apportioned (roughly in equal parts) between other recurrent expenses and drug purchasing (but drug expenditure may get a much higher share where private firms and brand medicines dominate the scene). If the total expenditure structure is found dramatically different from the described pattern, serious distortions are probably present and need to be addressed. For instance, the expenses related to security and logistic may absorb most available funding.

Step five. Identify the major flaws affecting the health sector and consider the realistic policy options available to decision makers (internal and external) to address such flaws, given the projected financing levels and the present situation in the health sector. The table presented later in this module includes some of the most common problems affecting health sectors emerging from a protracted crisis. Some of the policy options worth of consideration are sketched and commented. Annex 12 offers a tentative application of this conceptual approach to the Iraq health sector in 2003. Consider different service delivery models and service mixes (for a full discussion, see below, under Bottom-Up Approach, Second Round, and in Module 7).

Step six. Work out size and features of an affordable health sector, given existing constraints. Cost estimates of the sort described below, in the Bottom-Up approach, are needed to translate forecasted financing levels into number of facilities and health workers. Estimate the service coverage obtained by the health sector projected in the previous step. Consider the management systems needed to run the revamped health sector, according to proposed size, features and health care delivery model. Work out the legal, institutional and financial implications of implanting performing management systems. For a discussion of the issue, see Module 8.

Enlist the potential efficiency gains on offer and discuss the feasibility of measures aimed at achieving them. For instance, the introduction of a centralized mechanism to purchase generic drugs through international competitive bidding can boost drug availability through the health sector. Other savings, such as downsizing a bloated workforce through an aggressive retrenchment programme, can be politically much more difficult to enforce, particularly in a post-war environment.

Step seven. Consider the sustainability, balance, equity, efficiency, effectiveness of the projected health sector. Even if all these aspects have been considered earlier in the iteration, a fresh appraisal of the results attained is recommended. Identify the additional interventions deemed as necessary. Proceed to the needed adjustments. Consider whether the problems that would remain after revamping the health sector can be more effectively addressed by adopting an alternative service delivery model.

Common systemic flaws and possible policy responses [pdf 27kb]

Bottom-up approach

Recurrent and capital unit costs for health facilities of different levels of care provide the starting point. As international experience has shown that in the long term the financing of recurrent costs is likely to constitute the most serious constraint for health sector development (Segall, 1991), the total cost incurred in operating a recovered health care network is given a dominant weight in the reasoning proposed below. All health facilities, public and private (for-profit and not-for-profit), should be considered in the initial analysis. Later, each sub-sector can be studied in isolation.

First round: elaborating a reconstruction scenario, while maintaining the present service delivery model

Step one: Obtain / formulate average total recurrent unit costs for different categories of health facilities and for different level of performance (satisfactory, average and poor). Given that they are expensive, labour-intensive and require adequate expertise to yield reliable results, costing exercises (commissioned by government, NGOs, charities or firms) are usually carried out on small samples of facilities. Obviously, the resulting costs can be used only when the studied facilities are representative of larger groups. In some cases, no estimate is available and figures from other countries considered reasonably similar to the one under study can be adapted as temporary proxies.

In a disrupted health sector, ‘facility unit costs’, both recurrent and capital, can be a very vague concept. Most derelict, under-supplied and underused facilities, offering dismal levels of care, yield low running costs when they are studied. Conversely, overstaffed facilities in secure areas, supported by well-resourced NGOs, incur very high running costs. Remote facilities performing outreach activities are more expensive to run (per unit of output) than urban ones with heavy patient loads. Thus, considerable caution is demanded when considering available unit cost figures. The values eventually accepted for facilities of ‘satisfactory’ performance should correspond as much as possible to the costs of a ‘normalised’ situation, where wartime distortions are removed, service delivery is largely an indigenous responsibility and operational standards have attained acceptable levels. An additional difficulty is posed by the categorization of existing health facilities, which can be extremely heterogeneous, particularly when built and operated by NGOs. A ‘shadow’ functional classification of facilities can be needed to strengthen estimates (see Module 9).

Cost figures should include both expenditure incurred at the delivery point and that paid for elsewhere, but related to the service production process, such as drug or food donations or in-service training provided by a third party. As in troubled situations a large part of the inputs consumed in health care production are not accounted for in the formal budget of the involved facility, a detailed inventory of the absorbed inputs must be carried out locally. As a rule of thumb, the total recurrent costs of average facilities increase several times moving upwards from one level of care to the next. Thus, the recurrent expenditure of a first-referral hospital may be 2-5 times higher than that incurred by a health centre offering a comprehensive basic package of services. These ratios, once refined according to size of facilities, number of beds, staffing patterns etc., provide precious indications about the optimal facility mix of the future network. Planners may present interesting options to decision-makers, such as between building an additional rural hospital instead of three health centres, or the other way round. In this way, decision-making gains content and policies meaning.

True story #12: Estimating the cost of revamping the health network in Mozambique

In 1992, the Ministry of Health of Mozambique finalized a strategy for post-war reconstruction (Noormahomed and Segall, 1994), which set the broad features of the future, recovered and sustainable health sector. In relation to the health infrastructure, the strategy aimed at significantly expanding the number and scope of PHC facilities and first-referral hospitals, while rehabilitating without enlarging tertiary hospitals. According to the chosen approach, cost estimates elaborated at the time assigned 55% of total investment (projected at approximately US$ 280 million) to PHC.

In 1998, when reconstruction was under way, new cost estimates were elaborated and compared to the original ones. While the PHC investment fit fairly accurately into the forecasted one, hospital costs accounted for twice the originally planned investment. This cost escalation took place without any increase in the number of targeted hospitals. In the new estimates, the share of investment allocated to PHC was reduced to 30% of the total. This substantive change was not due to mistakes in the original computations or to a policy change, but resulted from hospital recovery plans developed in isolation from each other. Architects and hospital doctors connived, on perfectly reasonable grounds, in identifying additional technical needs for each benefited facility, whose satisfaction sent the cumulative cost of reconstruction well beyond the ceiling originally agreed and considered sustainable.

Step two: Estimate the size and composition of the existing network, trying to remove ‘ghost’ facilities, i.e., those destroyed or permanently closed down, from available records. The cleaning of the available data may present considerable difficulties. Sometimes, the reliability of data can be improved by triangulating lists elaborated by different departments, such as planning, human resources and supply. Where an Essential Drug Programme is in place, its data can be very helpful in discriminating active facilities from inactive ones. Also, special programmes keep often information related to their specific area. Thus, a list of facilities providing emergency obstetric care helps to identify hospitals providing surgical services. Usually, the several available lists compiled by different parties present striking inconsistencies. A way to handle them is to build a nominal database of health facilities, with which many problems can be spotted and reconciled, after patient enquiry with health authorities, NGOs or knowledgeable people. Each facility must be characterized by key variables (number of beds, staffing, functioning laboratory, fridge, vehicle etc.), selected in order to determine its functional capacity. Given the quick pace of change typical of transitional processes, such a network database needs continuous updating to remain useful (see Annex 9 for a discussion of the details).

Classify existing facilities, according to their performance level. Reliable inventories of the network are rarely available, so that a rough estimate can be obtained by consulting a panel of knowledgeable people. Compute the total cost of operating the existing network at the present level of performance. To obtain the total cost of the health sector, add support systems (administration, training, transport, warehousing etc.). In most cases, these additional costs fall in the order of 10-30% of the total expenditure incurred by direct service delivery. Verify that the total recurrent expenditure computed in this way roughly matches the figure estimated from a macro perspective. Reconcile discrepancies, if found. Consider present levels of coverage and consumption of basic services. Compare them to the existing health care network, staffing patterns, levels of supply etc. In many instances, reported resources will appear large in relation to reported outputs. Identify major shortcomings and inefficiencies in service delivery. A common pattern is the underutilisation of peripheral facilities, both of primary and secondary level, due to inadequate support, reduced access, poor performance.

Step three: Estimate the total recurrent costs induced by a revamped network (setting, for example, that 80% of existing facilities perform at a satisfactory level), without changing its size and structure. As better quality of care is likely to induce increased utilization of services, a forecast of the levels of coverage and consumption of basic services that would be attained by revamping the performance of such a network can be formulated. Try to estimate the impact on utilization of a change in the existing user fee policy. When the new policy entails regulating a widespread practice, without increasing the cost to users of accessing services, its impact can be favourable. Conversely, a net increase in the cost shouldered by users could trigger a contraction of consumption. Disaggregate by regions, provinces or states, so as to spot underprivileged situations. Compute the total investment needed to achieve a satisfactory performance for the present health care network, including the revamping of management systems. Identify and cost interventions aimed at removing existing bottlenecks. For example, if skilled staff are absent in peripheral health facilities, building houses for them and/or providing hardship salary supplements may result more effective than training additional cadres.

Step four: Estimate the potential savings obtained by identifying and correcting some self-evident major inefficiencies, such as by closing down or downgrading redundant facilities in over-served areas, by re-deploying staff, by downsizing the workforce, by improving supply systems. Assess the political cost of the implied measures and the likelihood that they are adopted. Compute again total recurrent costs and projected coverages / consumption in a system where efficiency has significantly improved.

Estimate the additional expenses induced by integrating into the present health care network facilities run by hostile parties, or dismissed by the military, if the terms of the peace agreement imply such devolution. In some cases, like in Southern Sudan in 2004, the burden of running hospitals located in garrison towns and operated by the central government might be heavy, to the point of changing the whole financial outlook of the emerging autonomous health sector.

Step five: Project the additional infrastructure needed to correct existing inequalities in basic service consumption (split by high-case, mid-case and low-case scenarios, if possible). For instance, if baseline basic services are estimated to cover about half of the population, the implications in terms of new facilities to be opened, of staffing and supplying them and of increased recurrent costs could be explored for coverage targets of 60%, 70% and 80% (for a discussion of the problems related to population figures, see Module 2 and 4). The respective computations should pay attention to the diminishing returns of expanding the access to basic services, hence the increasing marginal costs of the projected growth. This is largely due to uneven patterns of population settlements. The initial service expansion is likely to benefit densely populated and easily reached areas, where health care is provided at lower unit cost. Later phases will demand the coverage of remote areas with sparse populations, where costs increase dramatically. Thus, if the standard basic health centre is planned to serve an average of 20,000 people, a ratio of 15,000 and even 10,000 might be appropriate to cover low-density areas (depending on settlement patterns).

Pay special attention to temporary population settlements, whose coverage with fixed services bears the potential of permanently distorting the health network. This problem can reach serious dimensions in situations where displaced populations (internally and abroad) are large. IDPs, usually poorly known and contended for by warring parties, pose a more difficult challenge than refugees, whose formal status generates better information about their number, settlement and health status.

Project the additional burden the HIV/AIDS epidemic is likely to place on the health sector within the planning timeframe. For countries already badly hit, the disease alters health care demand and the response(s) to it. Most aspects of health care provision, likely to be affected, must be considered by the recovery strategy. The demand for inpatient care, laboratory services, drugs, skilled practitioners and nurses are all expected to increase substantially. The ability of people to pay for health care is correspondingly reduced. Increased dependence on external assistance is in most cases an inevitable outcome. For a brief discussion of the relationships of HIV/AIDS and complex political emergencies, see Annex 4b.

Step six: Work out the total recurrent costs incurred by the expanded / restructured network and the consequent gains in terms of coverages and consumption. Work out the implications / constraints of the proposed expansion in terms of human resource requirements, management systems etc. Try to estimate the incurred costs of transforming the health sector according to the newly set targets.

Step seven: Choose the alternative considered as affordable from a macro perspective (in terms of recurrent expenditure), according to the estimates of total available financing developed in the Top-Down Approach. Work out the investment needed to attain the projected levels of coverage / consumption, also including support sub-systems, such as warehousing, transport and training. Establish a timeframe for health sector recovery, according to the chosen option. Seriously disrupted health sectors need long cycles (10-20 years) of sustained efforts to recover. Plans tend to underestimate the time demanded for huge, systemic interventions to approach completion. Model the evolution of available financing over time. In many instances, aid flows expand dramatically in the immediate post-conflict years, to recede quickly later, when most of the planned investment kicks off (Collier, 2002). Practical ways to address this mismatch must be identified and negotiated with financiers.

Consider the feasibility of the chosen option, given existing implementing capacity. In very poor countries, capacity may be as scarce as resources, in such a way that humble recovery plans become inescapable. In certain situations, where abundant mineral wealth encourages financially ambitious choices, capacity may become the decisive criterion for decision. Unfortunately, a common symptom of poor capacity is the unawareness of it. The Angolan health sector has consistently been fraught by the oil-induced perception of future opulence and by its capacity shortage. Overambitious, never implemented plans have regularly ensued. For a discussion of capacity, see Module 8.

Investment in human resources is particularly important, as its outcome materialises slowly and sometimes in ways diverging from the anticipated. Additionally, training is expensive, labour-intensive, culture-bound and technically demanding. And the workforce, appropriate and productive or not, will anyway absorb the largest part of the future recurrent expenditure. See Module 10. Pay special attention to referral systems, whose costs (both capital and recurrent) are always substantial. Rural first-referral hospitals are among the health sector components that suffer most from disruption and whose recovery presents special difficulties. See Module 9.

Second round: Introducing an alternative service delivery model, or a mix of old and new models

Particularly when the country has gone through a prolonged period of disruption, which has screened it from international developments in health care provision, the prevailing delivery model(s) may be perceived as outdated, particularly by new rulers. Policy makers (outsiders as well as insiders) may be attracted by new approach(es), potential candidate(s) to replace the old one(s). For instance, the Kosovo health sector suffers from a heavy hospital bias, inspired by Soviet planning criteria, bias which needs to be corrected if a viable system is to be built. In other contexts, major common distortions may include service fragmentation along vertical lines, overemphasis on facility-based care, out-of-control privatisation or, conversely, over-reliance on public provision (or a mix of most of these distortions). Old patterns of service delivery must be compared to alternative ones. Clearly, an alternative service delivery model must appear very promising, offering clear advantages over the old one, to be worthy of consideration. Health care delivery models are discussed in Module 7. Analysing Patterns of Health Care Provision.

A special case is posed by severe, protracted crises, whereby health service provision has evolved to such a degree to rule out the resuscitation of old models. In Afghanistan, where the public sector has closed down for years, a dominant share of health services is provided by NGOs. The reintroduction of the centrally-planned and financed public provision of health services in these settings seems out of question, at least in the short term. National authorities, encouraged by influential aid agencies, have opted for the formal regulation of the field, through the contracting out of service delivery to private non-for-profit operators. See Annex 7 and Module 8 for more details, respectively on contracting out and on regulation.

Adopting an alternative service delivery model usually implies the introduction of new management systems, or substantive changes to existing ones. These implications must be made explicit as soon as possible, before the choice is made. The financial and political cost of equipping the health sector with management systems adapted to a new delivery model may be substantial, and must be adequately considered when alternatives are appraised.

To elaborate costed estimates of the adoption of an alternative service delivery model poses additional difficulties, because data related to a different approach may not be available. In this case, experience from abroad may help. Also, small-scale pilots may contribute useful information to strengthen the computations. Tightly monitored experimentation in limited settings appears advisable, before a new model is adopted nation-wide. Considerable caution is needed in adopting successful pilot models for countrywide implementation, as pilots are by definition privileged endeavours, bound by nature to succeed in a way or another. Going to scale is always a challenge of different order.

New estimates must be elaborated as soon as reliable data become available. Despite the difficulties of costing the adoption of alternative service delivery models, this sort of estimates is needed to inform the policy discussion, which in their absence risks being driven mainly by ideological arguments.

Follow the same steps sketched in the first round and eventually compare projected results. Identify the most promising model (or a mix of models).

Materialising the recovery strategy: common pitfalls

A thoroughly crafted recovery strategy needs to be disseminated, understood and incorporated into the plans of the most important stakeholders, to stand true chances of being followed. To ensure widespread support to the strategy, negotiation and communication skills are as important as technical ones. Furthermore, awareness among its supporters of the priority of the recovery strategy over other concerns is paramount.

The recovery strategy may stumble, fail or become distorted, due to a host of reasons.

Recommended reading

Bower H. Reconstructing Afghanistan’s health system: Are lessons being learned from the past? MSc. Dissertation. London School of Hygiene and Tropical Medicine. 2002.
A brilliant inquiry into the complexities of the Afghan health sector at a time of dramatic changes. Very perceptive discussion of policy making and coordination in an extremely disrupted context. The relevance and applicability of experiences from abroad to the Afghan situation is realistically appraised.

Macrae J. Zwi A. B. Birungi H. A Healthy Peace? Rehabilitation and Development of the Health Sector in a ‘Post’-Conflict Situation. The case of Uganda. London School of Hygiene and Tropical Medicine. 1994.
A classic report, groundbreaking and very influential. Several of the patterns shaping transitional situations are described and critically discussed. A synthesis of the report’s main themes is given by Macrae J. Zwi A. B. Gilson L. A Triple Burden for Health Sector Reform: ‘Post’-Conflict Rehabilitation in Uganda. Soc. Sci. Med. Vol. 42, No. 7, pp. 1095-1108. 1996.

Noormahomed A. R. Segall M. The Public Health Sector in Mozambique: A post-war strategy for rehabilitation and sustained development (Portuguese original, 1992; English translation, WHO 1994, Macroeconomics, health and development series; no. 14).
This reconstruction strategy, developed before the end of the war by the Ministry of Health of Mozambique, was published by WHO as ‘best practice’. One decade later, it still deserves this title. Resulting from three years of studies and discussions and largely conceived by insiders, this document set a clear resource constraint for health sector recovery, planning what was at the time considered affordable in the long term. Its influence on the reconstruction process was vast. If the reconstruction of the health sector resulted in a (qualified) success, it was also because many autonomous actors tried vigorously to materialise the vision laid down in this document. Despite its age, recommended reading to every practitioner involved in a recovery process.

- Full text [pdf 3.8Mb]


Berg E. Why Aren’t Aid Organizations Better Learners? EGDI 2000.

Collier P. Aid, Policy and Growth in Post-Conflict Countries. The World Bank. Conflict Prevention and Reconstruction Unit. Dissemination Notes Number 2. April 2002.

Conway T. Current Issues in Sector-wide Approaches for Health Development: Cambodia Case Study. WHO/GPE/00.2. WHO, Geneva. 2000.

Hay R. The ‘Fiscal Space’ for Publicly-Financed Health Care. Policy Brief Nr. 4. Oxford Policy Institute. 2003.

Lanjouw S. Macrae J. and Zwi A. Rehabilitating health services in Cambodia: the challenge of coordination in political emergencies. Health Policy and Planning 14(3): 229-242. 1999.

McPake B. and Mills A. What can we learn from international comparisons of health systems and health system reform? Bulletin of the World Health Organization. 78(6). pp. 811-820. 2000.

Macrae J. Rebuilding health systems: an overview of the dilemmas. WHO Health in Emergencies. June 2002: 13; 1-2. 2002.

Segall M. Health Sector Planning Led by Management of Recurrent Expenditure: An Agenda for Action Research. International Journal of Health Planning and Management. Volume 6, pp. 37-75. 1991.

Strong L. Policy Transfer in Post-Conflict Settings Performance-based Partnership Agreements in Afghanistan. MSc. Dissertation. London School of Hygiene and Tropical Medicine. 2003.

Van Brabant K. Organisational and Institutional Learning in the Humanitarian Sector. Opening the Dialogue. ODI. 1997.

World Health Organisation. Health in Emergencies. Countries in Transition. Issue no 3. June 2002.

Annex 12: Case studies

The case of Mozambique

The situation at the end of the war (1990-1992)

The chosen approach to reconstruction

The resources made available for reconstruction

About US$ 300 million invested in the reconstruction process over a decade by donor agencies and development banks. Main channels:

The constraints met

The long-term results

The lessons to be retained for other reconstruction processes

The case of Uganda

by Maurizio Murru

The situation at the end of the war (1986)

The chosen approach to reconstruction

The resources made available for reconstruction

The constraints met

The long-term results

The lessons to be retained for other reconstruction processes

The case of Cambodia

by Peter Hill

The situation at the end of the war (1993)

The chosen approach to reconstruction

The resources made available for reconstruction

The constraints met

The long-term results

The lessons to be retained for other reconstruction processes

[an error occurred while processing this directive]