THEME PAPERS

What can we learn from international comparisons of health systems and health system reform?

Enseignements à tirer des comparaisons internationales entre systèmes de santé et réformes de ces systèmes

¿Qué lecciones cabe extaer de las comparaciones internacionales de los sistemas de salud y de las reformas de los mismos?

Barbara McPake^I; Anne Mills^II

^ISenior Lecturer, Health Economics and Financing Programme, London School of Hygiene and Tropical Medicine, Keppel Street, London WC1E 7HT, England
^IIProfessor, Health Economics and Financing Programme, London School of Hygiene and Tropical Medicine, London, England

Correspondence

ABSTRACT

Most commonly, lessons derived from comparisons of international health sector reform can only be generalized in a limited way to similar countries. However, there is little guidance as to what constitutes "similarity" in this respect. We propose that a framework for assessing similarity could be derived from the performance of individual policies in different contexts, and from the cause and effect processes related to the policies. We demonstrate this process by considering research evidence in the "public–private mix", and propose variables for an initial framework that we believe determine private involvement in the public health sector. The most influential model of public leadership places the private role in a contracting framework. Research in countries that have adopted this model suggests an additional list of variables to add to the framework. The variables can be grouped under the headings "demand factors", "supply factors", and "strength of the public sector". These illustrate the nature of a framework that could emerge, and which would help countries aiming to learn from international experience.

Keywords: comparative study; contract services, case studies; health care reform; cross-cultural comparison; outcome and process assessment (health care).

RÉSUMÉ

Il est rare de pouvoir généraliser à d'autres pays, au moyen de comparaisons internationales, les enseignements tirés de la réforme du secteur sanitaire d'un pays donné: la spécificité est si grande qu'on ne pourra retenir au mieux ces enseignements que dans un contexte précis. Le plus souvent, cet exercice devra se limiter aux pays semblables et dépendra de l'existence de modèles sous-jacents des processus de réforme et de la compréhension de leur impact sur les objectifs politiques. Toutefois, une généralisation limitée et, plus largement, une comparaison internationale des réformes de la santé, nécessite un cadre empirique pour déterminer si les pays sont «plus semblables» ou «moins semblables». Le procédé que nous proposons dans cet article établit ce cadre à partir d'une série d'actions politiques individuelles, de leurs résultats dans diverses situations et des causes et effets probables qui entrent en jeu. Prises individuellement, les politiques ont des dimensions sanitaires qui se recoupent largement. Nous avons supposé qu'il s'agissait là de facteurs de similarité, bien que ce dernier point ne puisse être confirmé qu'à la fin du processus que nous abordons dans cet article.

Nous étudions les faits en rapport avec la «complémentarité entre le secteur public et le secteur privé» pour les soins de santé et proposons un cadre préliminaire permettant aux pays de classer leur système de santé. Nous pensons que plusieurs facteurs, liés au développement du pays, influent sur l'existence des prestateurs privés: niveau des revenus, importance de la classe moyenne, prestations sanitaires du secteur public, capacité de celui-ci à mener à bien des activités de réglementation et de surveillance et à négocier avec le secteur privé, niveau de développement du secteur privé (en dehors de la santé), niveau de formation. Il y a également d'autres facteurs, indépendants du niveau de développement, comme le nombre de médecins et les structures institutionnelles du secteur public résultant des «facteurs interdépendants».

L'examen des résultats des prestateurs privés fait apparaître une conclusion: si le secteur privé peut compléter les prestations assurées par le secteur public, il ne peut optimiser la contribution de ce dernier à l'état de santé général de la population. Cela signifie qu'il convient d'accorder une plus grande confiance aux modèles de réformes qui renforcent l'autorité publique dans ce secteur. L'instauration d'un cadre contractuel autour de cette autorité a constitué un modèle puissant à l'origine de changements dans de nombreux pays. Quelques-uns dont la Colombie, le Royaume-Uni et la Zambie ont adopté ce principe pour l'ensemble de leur système de santé mais, en règle générale, les pays l'ont appliqué au coup par coup, passant des contrats pour certains services lorsque cela semblait avantageux.

L'examen des travaux de recherche a fait apparaître des paramètres pouvant influer sur la mise en œuvre du modèle dans différents pays: variables déterminant la mesure dans laquelle la loi de la concurrence se fait ressentir; tendances du financement du secteur public pendant la période de mise en place de la réforme; investissements dans le secteur public; souplesse des systèmes bureaucratiques; pouvoir des syndicats dans le secteur public; caractéristiques culturelles des relations avec le marché. Il est possible de regrouper les paramètres susceptibles d'influer sur l'engagement privé dans le secteur public de la santé en «facteurs liés à la demande», «facteurs liés à l'offre» et «force du secteur public» (voir Tableau 1). Selon nous, ces éléments peuvent servir à établir un cadre de comparaison internationale des systèmes de santé et à améliorer ainsi les informations mises à la disposition des pays envisageant de réformer leur secteur de la santé.

RESUMEN

Las lecciones que pueden extraerse de las comparaciones internacionales de los sistemas de salud y de las reformas de los mismos rara vez pueden generalizarse de un país a otro o, en el mejor de los casos, son tan específicas que sólo pueden aplicarse en un determinado contexto. En la mayoría de los casos las conclusiones se pueden generalizar de forma limitada a países similares. Esas generalizaciones limitadas son posibles cuando existen modelos de los procesos de reforma y se comprenden las repercusiones en los objetivos de política. Sin embargo, toda generalización limitada, y en general toda comparación de reformas sanitarias a nivel internacional, requiere un marco empírico que permita determinar el grado de similitud de los países. En este artículo proponemos un procedimiento para crear un marco de esa naturaleza a partir de una serie de políticas concretas, de su desempeño en diferentes contextos, y de las causas y los efectos probablemente implicados. Probablemente se darán coincidencias entre los distintos marcos a medida que se evalúen nuevas políticas, pero eso sólo podrá confirmarse al final del proceso iniciado con este artículo.

Hemos analizado las pruebas científicas disponibles sobre la atención sanitaria «mixta pública-privada», y proponemos un marco preliminar mediante el que los países podrían clasificar sus sistemas de salud. Sugerimos que el desarrollo de los proveedores privados se ve influido por varios factores asociados al nivel de desarrollo de un país. Entre esos factores cabe citar los siguientes: nivel de ingresos; tamaño de la clase media; prestaciones de atención sanitaria por el sector público; capacidad del sector público para llevar a cabo actividades de reglamentación y vigilancia y para negociar con el sector privado; nivel de desarrollo del sector privado (no sanitario); y nivel de educación. Otros factores no tienen por qué estar correlacionados con el nivel de desarrollo, entre ellos el suministro de médicos y las estructuras institucionales del sector de la salud resultantes de la «dependencia diacrónica».

Una conclusión se desprende del análisis de los datos sobre el desempeño de los proveedores privados, y es que, si bien puede complementar la prestación pública de salud, el sector privado no puede maximizar su contribución a la salud de la población. Eso hace pensar que habría que tener más confianza en los modelos de reforma que refuerzan el liderazgo público del sector. Un modelo poderoso que ha influido en los cambios en muchos países sitúa el liderazgo público en el interior de un marco contractual. Algunos países, como Colombia, el Reino Unido y Zambia, han aplicado ese principio en todo el sistema de salud, pero la mayoría lo han aplicado de manera poco sistemática, subcontratando servicios cuando ello parecía ventajoso.

Analizando las actividades de investigación se identificaron también diversas variables que pueden influir en la aplicación del modelo en diferentes países. Entre ellas cabe citar las siguientes: variables que influyen en el grado de manifestación de las fuerzas competitivas; tendencias de las finanzas del sector público durante el periodo de implantación de las reformas; inversión en el sector público; flexibilidad de los sistemas burocráticos; poder de las organizaciones de trabajadores en el sector público; y aspectos culturales de las relaciones de mercado. Las variables con influencia potencial en la implicación del sector privado en el sector público de la salud pueden clasificarse en «factores de demanda», «factores de oferta» y «poder del sector público» (véase la tabla 1). Creemos que es posible emplear esos componentes en un marco de comparación de los sistemas de salud internacionales, y que dicho marco podría mejorar la información disponible en los países que prevén reformar sus sectores sanitarios.

Introduction

According to Rose (1), "lesson drawing" in public policy starts with a commonly experienced problem that forces policy-makers to depart from routine and search for answers. The process of lesson drawing involves the following: searching elsewhere, developing a model of how a programme operates, devising a new programme and prospectively evaluating transfer. International comparative research on health system reform aims to contribute to this process by doing the following:

– describing and clarifying problems, and determining the extent to which they are commonly experienced in more than one country;

– identifying the range of approaches used in different countries to address problems;

– developing a theoretical model of how an intervention reduces a problem;

– testing the model in different countries and understanding how the contexts of individual countries affect performance of the model;

– revising the model in light of the tests;

– retrospectively evaluating the transfer of the model to other settings.

Both Rose (1) and Marmor (2) identify two fallacies in discussions of international health comparisons.

• The search for a single best model, for example in health sector finance. This overlooks the importance of context — the best model in one setting need not be so in another.

• The belief that nothing can be learned from other contexts, because differences in context are always crucial.

In contrast, both Rose and Marmor argue that the range of conclusions from international health comparisons can be generalized to different degrees. We would place this view in the framework of Fig. 1. There are two extremes of conclusions in this model — the generalizable and the specific.

We propose a third category of conclusions capable of limited generalization to "more similar" countries. However, categorization of countries into "more similar" and "less similar" groups requires a considered and empirically informed process, which we refer to as a "framework" for international comparisons of health systems.

Such a framework can be constructed by developing, testing and revising models of cause and effect in health policy, and particularly by prospectively evaluating policy transfer. Frameworks may be developed for a number of individual policies, but it is likely that individual frameworks will substantially overlap, and after a series of policy measures have been considered few new lessons will emerge as novel individual policies are assessed. However, this view can only be assessed at the conclusion of the process we propose here. Understanding the impact individual policies have in different contexts would allow future policies to be set in context and be identified as determinant. Thus the process may prove capable of assessing policy interdependence as well as the dependence of policy on the exogenous context.

In this article, we explore the functions of comparative research in health system reform by developing a framework for comparing international health systems, using as an example research on the public–private mix for health care.

Public–private mix in health care: development of the debate

Arguments over the merits of private versus public health systems go back at least to the 1960s, when Lees (3) argued that health care "would appear to have no characteristics which differentiate it sharply from other goods in the market" and hence should be provided through market mechanisms. By the 1980s, Lees' arguments for privatizing health systems were reflected in publications of the World Bank and in the views of commentators on health systems in less developed countries (LDCs), who largely derived their views and ideologies from the western side of the Atlantic (4–8). The promoters of more private systems, such as found in the USA, emphasized the value of competition and the stronger incentives for efficient performance that are associated with private institutional forms. On the other hand, promoters of more public systems, such as those found in the United Kingdom, emphasized the "market failures" implicit in health markets (e.g. 9) and the inappropriateness of applying generalizations from other sectors (10).

This debate was largely terminated by Culyer's demonstration that the question of which health system was better could only be decided using empirical evidence (11). Despite this, and although a common problem had been identified (poorly performing health systems in LDCs) and a model had been proposed (private financing of health systems improves performance through incentives and competition), most commentators did not research and evidence the "initial model" but seemed to consider its operation to be self-evident.

Role of private providers in developing countries: recent research and debate

Initial research efforts on the public–private mix for health care in LDCs began to take place in the early 1990s. These efforts took the international comparative research process forward to the stage of testing the theoretical model with experiences in different countries. Evidence from these efforts indicated that private provisioning and financing already played a much greater role in health care than was suggested by the typical characterizations of LDC health systems as being dominated by the public sector. Even in countries with very low levels of private sector provision, individuals' out-of-pocket payment for health services often exceed one-third of the national health expenditure (12). The evidence also suggested that the private sector was not used exclusively by upper-income groups. Rather, a range of private providers catered for different socioeconomic groups, from unqualified "quacks" and drug sellers operating in poor rural and urban areas (13–15), to well-equipped hospitals in richer urban areas.

While it might appear that the performance of private providers was superior (for example, in meeting "felt need" or preferences), evidence of market failures abounded, particularly because of information asymmetry — the relative lack of technical knowledge among consumers. Private-for-profit providers seldom offered services of a public health nature, even services such as immunization. Such providers over-prescribe in general, but especially antibiotics and injectables, which are understood to be over-valued by patients from a technical perspective (16, 17). Also in countries where private involvement in hospital provision is extensive, unnecessary technology was often adopted (18). Furthermore, competition could not be relied upon to effectively discipline private sector health providers, even by imposing pressures for efficient production of services deemed most valuable by the population. In many LDC markets competition and even "contestability" are absent (19); and information asymmetry allows inefficient forms of competition to arise, such as "quality signalling" (20).

McPake argued several years ago that issues related to the role of private health providers varied according to two important dimensions that predicted the level and form of private health sector development (21). The first of these was the level of economic development, and within the group of countries generally included on the LDC list there were important distinctions between the very poor and the transitional, or middle-income, nations. This dimension is likely to function as a proxy for a number of characteristics that might better be considered separately: income levels; size of the middle class; the public sector care package; the capacity of the public sector to regulate, monitor and negotiate with the private sector; level of private (non-health sector) development; and education levels.

The second dimension was population density. Higher population densities are indicative of a more developed private sector because large numbers of even very poor people can provide a sufficient market for certain types of private sector providers. A good example of this can be seen in the proliferation of private doctors in the Karachi kachi-abadis (urban slums) (22).

We now propose other important country and health sector characteristics. First, human resource development policy has a crucial influence on the supply side of the market. An excess supply of doctors may have been key in producing the proliferation of doctors in the Karachi slums, and this situation also exists elsewhere in the subcontinent; for example, in India (23). Second, institutional structures in the health sector have historical determinants unrelated to current development dynamics, although it may be difficult to isolate these structures purely as cause since they are also in part effect. For example, the institutional structures in countries with similar income, population and development (e.g. Bolivia and Côte d'Ivoire) have been moulded by different colonial legacies and cultural influences. In Bolivia, the institutional structure is tripartite and involves large public, social security, and private subsectors. In Côte d'Ivoire it is dominated by the public sector with only small-scale private sector activity (which nevertheless accounts for a substantial share of health expenditure). These differences alone seem to create differences in issues surrounding the role of private providers. Indeed, historical factors powerfully influence current structures and arrangements, as well as the extent to which they can be readily changed by deliberate policy (24), a phenomenon termed "path dependence" (25). In other words, institutions change only gradually, not least because of the power structures and personal relationships they embody.

Private sector participation in a public health sector framework

Despite differences in specific public–private issues, there is one common conclusion: although the private sector can complement public health provision and provide some types of services better, it cannot lead the health sector in a direction likely to maximize its contribution to the health of the population. This can be seen both in the role of private drug sellers in rural Ugandan villages and in that of exclusive private hospitals in Bangkok. Recognition of the need for public-interest, rather than profit-oriented, leadership in the health sector led to increased interest in a revised model for LDCs, that was already implicit in some industrialized country reforms (e.g. in Sweden and the United Kingdom).

According to the revised model, private providers and competitive pressures play a role within a guiding framework imposed by public health authorities. The guiding framework could take the form of regulation, but research has shown regulatory influence to be weak in most LDCs (26, 27). This model has an overlapping rationale with that of the initial model proposed by Lees (3). However, it emphasises the value of competition and the incentive structures of private organizations as spurs to good performance, while recognizing the need for a public role in resource allocation.

An alternative version of the revised model proposed that gains from improved management could be obtained within a contracting framework, without introducing either competitive pressures or private sector organizations. In addition, it has been argued that using contracts between funders and providers — even when both are public and there is no competition — allows funders to distance themselves from providers and focus instead on resource allocation. Consequently, it is argued that the contract is in itself a useful tool for changing the behaviour of providers (28).

There are therefore three overlapping rationales for introducing contract-based models: the value of competition and the incentive structures associated with private organizations; the importance of a publicly set guiding framework for resource allocation; and the belief that the contracting process is sufficient to induce performance improvements by itself. Reforms in a few countries have led to the almost wholesale adoption of contract-based models within a publicly set framework (e.g. Colombia and Zambia). And many other countries have plans for increasing contractual relationships, both within the public sector and between public and private sectors (29).

International comparative research is beginning to permit tentative conclusions to be drawn about the model's performance. In practice, however, it appears difficult to achieve the theoretical advantages of the contract model. The difficulties vary according to a range of health sector and other characteristics, many of which are correlated with a country's level of development. This can be seen by comparing three countries that adopted a more wholesale version of the model (Colombia, the United Kingdom, and Zambia), as well as from a review of the contractual relationships between public and private sectors for support services, such as catering and cleaning, in five Asian and African countries.

Transforming integrated health systems into contracting health systems: experiences in three countries

United Kingdom. In 1991 in the United Kingdom, a "purchaser–provider split" was introduced into an archetypal public integrated model. The main rationales appear to have been the creation of competition between hospitals and the development of a public purchasing role independent of providers. The expectations of the public purchasing role exemplify the arguments of the "revised" model. Public purchasers would be able to identify the needs of the population and distribute resources accordingly, in contrast to pure private markets.

In contradiction of the "natural monopoly" argument (30), a 1997 study found that competition was far from absent in much of the United Kingdom hospital market (31). However, political decisions still impeded competitive forces. For example, in London, where the need for hospital closure had been long recognized, political mandate rather than market forces determined which hospitals would close (32, 33). A second constraint to competitive forces arose from the demand side of the market. For example, it might be expected that a monopsony would strengthen public purchasers. However, in practice it caused conflict between seeking "best value" services in the short term, and protecting the long-term interests of a district population by ensuring the stability and survival of providers. The district health authority purchasers who controlled the largest share of resources were unable to ignore the interests of the district hospitals on which their population relied. The 1997 changes to National Health Service (NHS) structures seem to have recognized that the purchasing role needed changing, and relocated that role to primary care groups accountable to district health authorities, in which general practitioners (GPs), community nurses, and other interested parties participate. The creation of primary care groups may prove an important, albeit unintended, mechanism for strengthening competitive pressures.

The 1991 reforms did not fundamentally change reimbursement procedures at the primary level. GPs, for example, have always been independent (private), and yet have been largely publicly financed since 1948. The major change brought about by the 1991 reforms was to offer GPs the opportunity to replace the district health authority in purchasing certain hospital services. Those who did so became "fund-holding GPs". Because they purchased on a smaller scale than district authorities, they did not face the same constraints. For example, they could move contracts from one hospital to another without significantly affecting hospital viability. As a result, what competition emerged was for GP contracts. There were allegations that patients funded by GP contracts received preferential treatment, including prioritization on waiting lists (34, 35).

Colombia. In Colombia, a classic tripartite health system operated up to 1996. The reforms seem to have had three overriding objectives: to increase the solidarity of the health system by merging public and social security subsectors; to introduce competition; and to introduce a contracting environment within a public health framework. In pursuit of the first objective, public health budgets were redirected to subsidize social insurance for those sections of the population judged to be unable to pay (according to a near-universal household survey). This constitutes the "subsidized regimen" of the system.

To introduce competition on both provision and purchasing sides, health promotion enterprises (EPSs) were formed. The EPSs offered packages of care for the insured, under both subsidized and contributory regimens, by contracting with providers. The EPSs are private enterprises, which may or may not be for-profit and are sometimes community-based organizations. The latter, in particular, have competed for members of the subsidized regimen. This contract arrangement replaced integrated arrangements in both public and social security sectors.

Even though both purchasers and providers are essentially private, several factors move the system in a public health direction. First, extensive regulations set detailed rules of operation. Second, the market power of public funds flowing through the system can be leveraged. And finally, an "equalization mechanism" can enforce solidarity despite the proliferation of EPSs. Important rules of operation for the market include the following: a minimum package of care which all EPSs must offer; setting contribution and EPS reimbursement levels, for both contributory and subsidized regimens; and mandating a cross-subsidy from the contributory to the subsidized regimen. Over time, the plan is to equalize packages of care in the subsidized and contributory regimens.

Implementation of the reforms has required that public hospitals switch, with little preparation, from an integrated to a contract mode of operation, and from a budgeted financing system to one which bills different EPSs for services received by their members. Ensuring compliance with compulsory contributions by the self-employed is proving difficult, and planned flows of finance around the system have not always been realized. This has created windfall surpluses in some EPSs and financial crisis in other hospitals, despite a hugely increased volume of financing to the system as a whole. Within the contributory regimen, subscriptions (the bulk of which are still gathered by employers) have not always been transferred according to the wishes of employees. This contributes to "persistence", the tendency of the insured to stay in one fund despite freedom of choice to move to another (36, 37), but other causes may also apply. For example, those who change funds must incur transition and information costs, often including the need to change providers and establish new relationships and clinical histories. Perhaps for similar reasons, many of those who are newly entitled to insurance benefits do not actively select an EPS and continue to seek care in the public hospitals they have always used (38).

Zambia. In Zambia a purchasing agency, the Central Board of Health, has been set up at central level, separate from the Ministry of Health, but conceived of as the Ministry's "implementing agency". The Board agrees contracts with districts and purchases services from them for primary level and district hospitals. It also contracts for services directly with secondary and tertiary hospitals, which are now governed by hospital management boards, rather than by the Ministry. The Board also agrees contracts with nongovernmental hospitals and will shortly develop contracts with two hospitals previously owned by Zambia Consolidated Copper Mines, which are being divested as part of the privatization programme.

A number of problems have arisen, some because the reform package has been only partially legislated and ratified. According to a financing policy agreed by committee, but not yet ratified, the Central Board of Health should purchase a "basic package of care" and full cost-recovery fees should be charged for services outside the basic package. The basic package has been agreed upon for the district level, but is still under discussion for secondary and tertiary hospitals. The autonomy of hospital management boards has been heavily circumscribed by direct ministerial intervention, under a minister whose commitment to the reform programme was strongly doubted. This minister has recently been replaced and this may permit these problems to be addressed.

Other problems arise from what might be seen as an inadequately ambitious reform design. Contracts do not directly pay for service, but rather pay on the basis of agreed budgets and agreed outputs — similar to a "block contract". For example, secondary and tertiary hospital financing is based on the number of approved beds in each hospital, rather than on the volume of activity. Difficulties in effectively monitoring activity levels seem to preclude more sophisticated contract forms, but without these, it can be assumed that central financing is independent of performance by districts and hospitals, and there is little leverage exercised by the contracts themselves. What leverage might be implied by the reporting requirements of the contracts, which should demonstrate that agreed services have in fact been delivered, is undermined by the usual government failure to pay the full agreed budget because of revenue shortfalls. It is also not surprising that hospitals prefer to generate user charge revenue, irrespective of its effects on the core patient business intended to be subsidized by the Central Board of Health. In particular, it has been alleged that hospitals are redirecting resources into "high-cost" wards and other services which attract "cost-recovery" fees.

More positively, there has been progress in the development of a management information system. Financial reporting and hospital and district planning have clearly been developed in response to the reform agenda, and might not have met with the same success in the absence of the imperatives of the reformed system.

Finally, the performance of the reform programme cannot be judged without an understanding of the overall financial crisis of the health sector. Delays in the mine privatization programme ensured that the long-term declines in copper revenues have not been reversed, despite some rallying in the price of copper — the Zambian government's principal source of revenue. Following the introduction of health reforms, health sector revenues suffered substantial real cuts and the need to take over two of the mining company's hospitals, in towns which have no other facility, will spread resources still more thinly (39).

Conclusions. Maynard & Bloor's distinction between "regulated competition" and "managed competition" is useful in contrasting the Colombian and United Kingdom reform programmes (40). The authors define regulated competition as "competition on the supply side only, with maintenance of a single source of public finance" and managed competition as "competition on both the supply and funding (demand) sides of the market". According to these definitions, the United Kingdom reforms introduced regulated competition, whereas the Colombian reforms introduce managed competition. In Zambia, there is no competition on either side, but rather a system of public contracting. The common component in all three countries was the introduction of public sector contracts.

The three cases illustrate the general argument that it is difficult to achieve the theoretical advantages of contracting. In the United Kingdom, for example, competition was not absent, yet competitive forces were muted and perverse incentives arose to prioritize marginal business. In Colombia, the issue of "persistence", or perhaps the absence of genuine choice of insurer, may have been the most important block to competitive forces. In Zambia, reform design did not introduce any competitive forces. Other issues are likely to emerge as implementation proceeds.

Success in imposing a publicly set framework for health providers has also been mixed. In the United Kingdom, purchasing was constrained by the monopsonist character of district health authorities and it remains to be seen whether the switch to primary care groups will allow stronger public purchasing. In Colombia, the publicly set framework involved a regulatory rather than purchasing role. The failure to enforce regulations is already apparent, a predictable issue given the difficulties of regulation and the limited extent of public sector capacity. In Zambia, several missed opportunities for reform imply a failure to impose a public agenda on provider institutions. Besides adequate public sector capacity, the ability to impose a publicly set agenda also appears to require adequate funding of reforms. Using contracts as the basis of purchaser–provider relationships appears more costly than direct management (33, 41) In both Colombia and the United Kingdom, reform was accompanied by increased financing, whereas in Zambia continued decline prevented the effective imposition of the public agenda. Provider institutions who have not received agreed payments do not feel bound by contract agreements, and pursue alternative incentives provided by "private" patients.

The use of contracts to stimulate improvement of management systems is the most convincing rationale for contracting policy, and the evidence above supports this argument. Improved management information systems are apparent in all three countries and the reforms seem at least partially responsible.

Contracts between public and private sectors for support services

Far more common than the wholesale adoption of a contracting framework at the systems level has been the partial use of contracts to resolve specific subsector problems. A study of the implementation of "new public management" approaches in Ghana, India, Sri Lanka, Thailand, and Zimbabwe, countries at different levels of development, provides an opportunity to compare the problems faced by this approach in different types of countries (24). Since the nature of contracts and the difficulties they face are substantially different, depending on the nature of the service contracted for (42), the discussion here is confined to contracts for hospital-support services, such as catering and cleaning.

Ghana. Ghana had had a long-standing experience of agreements with church providers, but had made little progress in contracting out support services, despite intentions to do so (43). Ministry of Health staff lacked the skills and experience to design contracts, and there was concern that the proposals would encounter strong staff and union opposition. In addition, the level of private sector development was a serious constraint, with few firms available or willing to take on government contracts. Moreover, private sector wages were higher and limited the scope for the contracted service to be less costly.

Zimbabwe. In contrast, in recent years Zimbabwe has managed both to plan and implement a policy for contracting support services in the main hospitals (24, 44). Key factors which helped to account for this include the following: a carefully planned process of capacity building among Ministry of Health staff; arrangements to help affected staff either get jobs with the contractors or set up their own companies to bid for contracts; strong Cabinet backing for the policy; a more developed private sector with greater capacity to tender for contracts; and a contracting process which encouraged private sector bids for what might be regarded as a risky venture.

India and Sri Lanka. Both India and Sri Lanka had widespread contracting of support services such as catering and cleaning, especially in the larger cities (45–47). In contrast to Africa, availability of private sector suppliers was not a major problem; neither was the level of costs in the private sector; nor was the ability of the government administration to design and manage contracts. However, neither country had a national policy for contracting, and contracting arrangements were on an ad hoc basis. Constraints to greater use of contracts were more political than administrative. In both India and Sri Lanka a clear constraint was the inability to retrench public sector workers and address the power of the public service unions in dictating employment terms and conditions. Hence institutional factors, and especially the long-standing influence of a dominant public-sector ideology, were major barriers. In addition, although private-sector capacity existed to take on contracts, there were concerns that markets were not very competitive, both because of action by suppliers to limit competition, and because government prices and procedures discouraged substantial numbers of bidders.

Thailand. Thailand had quite extensive contracting of support services, carried out as a hospital initiative within strict central regulations (48). As part of the plans to restrict the growth of the civil service, contracting out was widely encouraged and regulations were relaxed. The level of private sector development also permitted extensive contracting. In contrast, there were problems on the government side, with regulations that did not encourage sufficient competition for contracts, or leverage government purchasing power to obtain favourable terms.

This brief summary highlights the extent to which progress and problems were dependent on some key country features. Government skills and experience were constraints in the African context, where Ministries of Health had not been accustomed to managing contractual relationships for services. However, the experience in Zimbabwe demonstrated that such constraints can be addressed though efforts to increase skills and provide relevant experience; also skills can be developed "on the job". More fundamental constraints to successful contracting out were external to the Ministry and included: unreformed and centralized bureaucratic systems that did not provide flexibility in how contracts were specified and managed, or give managers much authority or responsibility; an inability to address the labour implications of contracting; and a lack of political backing to pursue courses unpopular with public sector workers. A contrast can be drawn with the United Kingdom, where a government-wide policy of contracting out was implemented in the 1980s (prior to the specific health sector reforms discussed above). Factors influencing its success included concurrent legislation which weakened the power of organized labour, and the low status of unskilled service workers. Public sector workers in developing countries, in contrast, are generally in a more privileged position.

However, the context and capacity of the purchaser only partly explain the different progress between the countries. Equally important was the degree of development of the private sector, and especially the extent to which market relationships predominated. Market-based relationships, such as those which emphasize competition, imply very different sorts of interactions (49): competitive bidding requires firms who do not coordinate with each other over the bidding process; there should be no exchange of gifts in the bidding process between governments and providers; and monitoring of contracts needs to be backed by law. Changing the nature of the relationships between purchasers and providers is likely to be a long-term process.

Towards improving comparative analysis of health sector reform

This article has focused on public and private roles in the health sector, in order to explore the process of international health reform in a specific context. A comparison of the conclusions suggest that a few are generalizable, while others are highly specific. However, most of the useful output from the review can be classified as "categorizable conclusions": in different circumstances, different outcomes can be predicted from similar policies. This is not a very surprising finding, and indeed echoes the views of Rose and Marmor (1, 2). We have argued that as international comparative research accumulates, models of the relationships between policy variables and outcomes are constructed and amended. These models facilitate comprehension of critical variables that influence policy outcomes. In doing so, they suggest a framework for the comparison of countries and their health systems, and the similarity of experience and outcome with health sector reforms can be predicted.

However, little guidance is available to countries that would enable them to distinguish between instructive and misleading reform experiences. There is a considerable literature on the comparative analysis of health systems (e.g. see review 50), but it is too broad to enable a focused enquiry into the relevance and transferability of reform experiences. In most of this work, too, developing countries have been excluded, or included only as an afterthought (51). This has led most thinking on the subject to be largely intuitive. It is usually the case, for example, that discussions of international health sector reform consider either "industrialized" or "developing" countries; and on rare occasions, conclusions from such discussions are categorized along the dimension of development. For example, The World Bank's World development report 1993 (52) included a table in which the policy implications were categorized for countries at three different levels of development.

The conclusions from our review of public–private issues support the view that development levels are important when making comparisons of reform experience. Other important factors are given in Table 1. Many of these variables will be correlated with indicators of development. However, others are not very closely correlated (e.g. the culture of market relationships), and some not at all (e.g. labour market disequilibria). Together they enable a more specific interpretation of the vague concept "level of development" for those seeking guidance for international health sector reform.

Perhaps more importantly, frameworks such as those developed in Table 1 permit a more sophisticated use of experiences with international health sector reform that extends beyond considerations of transferability, per se. Using such frameworks and models that predict the importance of included variables, it is possible to propose mechanisms through which policy affects outcomes and identify potential constraints to desired outcomes. Hence, appropriate preparatory and contemporaneous measures can be developed and form part of a well-designed health reform package. It is here that international comparative research is likely to be most useful.

In this article we have attempted to provide the first steps in a process that we believe could enable a more coherent approach to international "lesson drawing" in health sector reform than is possible at present. The gap between the status quo and a coherent and comprehensive framework is considerable, but we believe that efforts to close it are well worthwhile.

Acknowledgements

The Health Economics and Financing Programme receives a research programme grant from the United Kingdom Department for International Development (DFID). The empirical research reported here is drawn from studies funded by the International Clearing House for Health Sector Reform Initiatives (Colombia), DFID HP Accord (Zambia), and DFID ESCOR (contracting studies).

References

1. Rose R. Lesson drawing in public policy: a guide to learning across time and space. Chatam, NJ, Chatham House Publishers, 1993.         

2. Marmor T. Global health policy reform: misleading mythology or learning opportunity, in: Altenstetter C, Bjorkman JW, eds. Health policy reform, national variations and globalisation. Basingstoke, Macmillan, 1997.         

3. Lees DS. Health through choice. London, Institute of Economic Affairs, 1961 (Hobart Paper No. 14).         

4. Akin J, Birdsall N, de Ferranti D. Financing health services in developing countries: an agenda for reform. Washington, DC, The World Bank, 1987.         

5. Lewis M, Miller L. Public-private partnership in water supply and sanitation in sub-Saharan Africa. Health Policy and Planning, 1987, 2 (1): 70–79.         

6. Roth G. The private provision of public services in developing countries. Oxford, Oxford University Press, 1988 (EDI Series in Economic Development).         

7. Birdsall N. Thoughts on good health and good government. Daedalus, 1989, 118: 89–123.        

8. Griffin CC. Strengthening the health services in developing countries through the private sector. Washington, DC, International Finance Corporation, 1989 (Discussion Paper 4).         

9. Arrow KJ. Uncertainty and the welfare economics of medical care. American Economic Review, 1963, 53 (5): 941–973.         

10. Mills A. Leopard or chameleon? The changing character of international health economics. Tropical Medicine and International Health, 1997, 2 (10): 963–977.         

11. Culyer AJ. The nature of the commodity 'health care' and its efficient allocation. Oxford Economic Papers, 1971, 23 (2): 189–211.         

12. Berman P et al. Private and non-government providers: partners for public health in Africa. Boston, MA, Harvard School of Public Health, Department of Population and International Health, 1995 (Conference report. Data for Decision-making Project).         

13. Claquin P. Private health care providers in rural Bangladesh. Social Science and Medicine, 1981, 15 (B): 153–157.        

14. Whyte SR. Medicines and self-help. The privatisation of health care in Eastern Uganda. In: Hansen H, Twaddle M, eds. Changing Uganda: the dilemma of structural adjustment and revolutionary change. London, James Currey, 1990: 130–148.         

15. Birunji H. Injection use and practices in Uganda. Geneva, World Health Organization, 1994 (unpublished document WHO/DAP/94.18).         

16. Aljunid S, Zwi A. Public and private practitioners in a rural district of Malaysia: complements or substitutes? In: Bennett S, McPake B, Mills A, eds. Private health providers in developing countries: serving the public interest? London, Zed Books, 1997: 40–53.         

17. Gilson L. Value for money? The efficiency of primary health care units in Tanzania. London, University of London, 1992 (PhD thesis).         

18. Nitayarumphong S, Tangcharoensathien V. Thailand: private health care out of control? Health Policy and Planning, 1994, 9 (1): 31–41.         

19. Bennett, S. Health care markets: defining characteristics. In: Bennett S, McPake B, Mills A, eds. Private health providers in developing countries: serving the public interest? London, Zed Books,1997: 85–101.         

20. Bennett S. The nature of competition among private hospitals in Bangkok. In: Bennett S, McPake B, Mills A, eds. Private health providers in developing countries: serving the public interest? London, Zed Books, 1997: 102–123.         

21. McPake B. The role of the private sector in health service provision. In: Bennett S, McPake B, Mills A. Private health providers in developing countries: serving the public interest?, London, Zed Books, 1997: 21–39.         

22. Thaver I. et al. Private practitioners in the slums of Karachi: what quality of care do they offer? Social Science and Medicine, 1998, 46 (11): 1441–1449.         

23. Yesudian CAK. Behaviour of the private sector in the health market of Bombay. Health Policy and Planning, 1994, 9 (1): 72–80.         

24. Mills A et al. The challenge of health sector reform: what must governments do? Basingstoke, Macmillan (in press).         

25. North DC. Institutions, institutional change and economic performance. Cambridge, Cambridge University Press, 1990: 3–10.         

26. Kumaranayake L. Economic aspects of health sector regulation: strategic choices for low and middle income countries. London, London School of Hygiene and Tropical Medicine, Department of Public Health and Policy, Health Policy Unit, 1998.         

27. Kumaranayake L, Hongoro C. Health sector regulation in Zimbabwe. London, London School of Hygiene and Tropical Medicine, Department of Public Health and Policy Health Policy Unit (forthcoming).         

28. Mills A. Improving the efficiency of public sector health services in developing countries: bureaucratic versus market approaches. In: Colclough C, ed. Marketizing education and health in developing countries. Oxford, Clarendon Press, 1997: 33.         

29. Mills A. Reforming health sectors. In: Mills A, ed. Reforming health sectors: fashions, passions and common sense. London, Kegan Paul (in press).         

30. Culyer AJ, Posnett JW. Hospital behaviour and competition. In: Culyer AJ, Maynard AK, Posnett JW, eds. Competition in health care: reforming the NHS. Basingstoke, Macmillan, 1990: 12–47.         

31. Propper C, Bartlett W. The impact of competition on the behaviour of NHS trusts. In: Flynn R, Williams G, eds. Contracting for health: quasi-markets and the National Health Service. Oxford, Oxford University Press, 1997: 115–134.         

32. Propper C. Agency and incentives in the NHS internal market. Social Science and Medicine, 1995, 40: 1683–1690.         

33. Ham C. Management and competition in the NHS, 2nd edit . Abingdon, Radcliffe Medical Press, 1997.         

34. West PA. Understanding the NHS reforms: the creation of incentives? Buckingham, Open University Press, 1997.        

35. Le Grand J, Mays N, Mulligan J. Learning from the NHS internal market: a review of the evidence. London, King's Fund, 1998.         

36. Bertranou FM. Are market oriented health insurance reforms possible in Latin America? The cases of Argentina, Chile and Colombia. Health Policy, 1999, 47: 19–36.         

37. Neipp J, Zeckhauser R. Persistence in the choice of health plans. In: Scheffler RM, Rossiter LE, eds. Advances in health economics and health services research, vol. 6. New York, Academic Press, 1985: 47–72.         

38. Yepes F, Sanchez LH. Reforming pluralist systems: the case of Colombia. In: Mills A, ed. Reforming health sectors: fashions, passions and common sense. London, Kegan Paul (in press).         

39. Kamwanga J et al. Autonomous hospitals in Zambia and the equity implications of the market for hospital services. London, London School of Hygiene and Tropical Medicine (unpublished report).         

40. Maynard A, Bloor K. Health care reform: informing difficult choices. International Journal of Health Planning and Management, 1995, 10: 247–264.         

41. Mills A. Contractual relationships between government and the commercial private sector in developing countries. In: Bennett S, McPake B, Mills A, eds. Private health providers in developing countries: serving the public interest? London, Zed Books, 1997: 189–213.         

42. Bennett S, Mills A. Government capacity to contract: health sector experience and lessons. Public Administration and Development, 1998, 18: 307–326.         

43. Smithson P, Asamoa-Baah A, Mills A. The case of the health sector in Ghana. Birmingham, University of Birmingham, Development Administration Group, 1997 (The Role of Government in Adjusting Economies, Paper No. 26).         

44. Russell S et al. Reforming the health sector in Zimbabwe: does government have the capacity? University of Birmingham, Development Administration Group, 1997 (The Role of Government in Adjusting Economies, Paper No. 20).         

45. Bhatia M, Mills A. The contracting-out of dietary services by public hospitals in Bombay. In: Bennett S, McPake B, Mills A, eds. Private health providers in developing countries: serving the public interest? London, Zed Books, 1997: 250–263.         

46. Bennett S, Muraleedharan VR. Reforming the role of government in the Tamil Nadu health sector. Birmingham, University of Birmingham, Development Administration Group, 1997 (The Role of Government in Adjusting Economies, Paper No. 28).         

47.Russell S, Attanayake N. Reforming the health sector in Sri Lanka: does government have the capacity? Birmingham, University of Birmingham, Development Administration Group, (The Role of Government in Adjusting Economies, Paper No. 14).         

48. Bennett S et al. The health sector in Thailand. University of Birmingham, Development Administration Group, 1997 (The Role of Government in Adjusting Economies, Paper No. 31).         

49. Flynn N. Comments. Paper presented at: Conference on the role of government in adjusting economies, Birmingham, University of Birmingham, 28–30 March 1999.         

50. Elling RH. Cross-national study of health systems: concepts, methods and data sources (vol. II); and Countries, world regions and special problems (vol. III). Detroit, MI, Gale Research Company, 1980.         

51. McPake B, Machray C. International comparisons of health sector reform: towards a comparative framework for developing countries. Journal of International Development, 1997, 9 (4): 621–630.         

52. World Bank. World development report 1993: investing in health. New York, Oxford University Press for The World Bank, 1993.        

Correspondence
Barbara McPake
Health Economics and Financing Programme, London School of Hygiene and Tropical Medicine
Keppel Street, London WC1E 7HT, England
E-mail: b.mcpake@lshtm.ac.uk