The World Health Report 2000: 10 years on

The publication of the World Health Report 2000 was an act of remarkable courage (WHO 2000). The World Health Organization (WHO) is extremely sensitive to complaints from health ministers in its member states and its legal department goes to enormous lengths to avoid anything being said that might be seen as critical even, as this author has experienced, of countries such as the USSR that no longer exist. Thus, few observers thought it was possible that the then Director-General, Gro Harlem Brundtland, would sanction the production of a ranking of the world’s health systems. Yet it did and the response was predictable. Those governments that scored highly, such as France, were delighted, while others, such as Brazil, that came much lower, formally expressed their displeasure at the WHO Executive Board (WHO 2001). In this commentary I focus on the rankings of health system performance contained in the report’s statistical annex. The framework for understanding health systems employed in the main text has proven uncontroversial and is now used widely, while the accompanying text is an extremely valuable source of material for scholars of health systems. In contrast, the country rankings have attracted considerable comment from researchers and politicians, much of it critical. Some, such as Navarro (2001), focused on what they perceived as an underlying pro-market ideology in many of the solutions proposed and the language used to justify them, in particular a seeming conflation of tax-funded national health services with the discredited Soviet system (perhaps anticipating the neo-liberal critiques of President Obama’s health care reforms a decade later). He and others (Almeida et al.2001) criticized what they saw as an unjustified dismissal of the primary care model set out at Alma Ata, which they attribute to the authors’ ideological opposition to an active role of government in funding and delivering health care. However, this view was not universal, with one commentator, writing in the Wall Street Journal, accusing the WHO of adopting a Marxist stance by arguing for any more than a token role for the state in providing health care (Helms 2000). Reflecting this diversity, Williams (2001) asked whether there are, in fact, universally agreed goals for health systems. Navarro (2000) also expressed concern about the face validity of the rankings, identifying countries, such as Spain and Italy, which scored highly on performance yet fared poorly in satisfaction surveys among their populations. Similarly, ranking Colombia higher on fairness of financing than Canada caused some surprise. Others expressing concern about face validity included Williams (2001), who expressed surprise that Denmark, with universal coverage, ranked lower on responsiveness than the USA. One paper noted the striking similarity between the overall health system rankings and those of their national football teams, as produced by FIFA (Appleby and Street 2001). The main criticisms were, however, of technical aspects of the methods used to assess performance. By common consent those undertaking this exercise faced a major problem if they were to include all 192 WHO Member States, some of which barely functioned as states and certainly lacked control over all of their territory. Many lacked even the most basic of information. Only a minority had any functioning system of vital registration, so if measures such as health outcomes were to be included they had to be modelled. This would be complicated enough if only the simplest of measures, such as life expectancy, was used. In many countries, there were established methods to calculate this by applying data from surveys of child mortality to standard life tables, but both inputs incorporated many assumptions. Yet a more complex measure, Disability Adjusted Life Years (Murray et al. 2000; Mathers et al.2001) was used, even though even fewer data on disability were available and their application required highly contentious assumptions about valuations of different health states (Anand and Hanson 1997). The estimates of distribution of health outcomes also attracted criticism; derived from data from only a few countries and involving complex modelling, the measure was essentially the sum of intra-individual differences in outcome. This takes no account of the social and ethnic patterning of health whereby a summary measure may designate a system as reasonably fair even though there was systematic discrimination against a minority (Braveman et al.2000). The problems in assessing responsiveness were even greater, with data being modelled on the basis of results from a survey of 1791 respondents in 35 countries, many of whom were WHO staff (Williams 2001). The modelling process did exclude those results for which ‘no rational explanation’ could be found and adjusted them for the tendency of those living in dictatorships to score their systems especially highly, but, even after this process, others showed poor correlation between the opinions of key informants and those surveys that existed (Blendon et al.2001). Other criticisms were levelled at the measures of fairness of financing, for not taking account of the greater needs of the poor or the scale of redistribution, the use of a composite index of quite different parameters whose meaning was difficult to visualise, and the lack of external peer review (Almeida et al.2001). The WHO has responded to these criticisms, for example by asking whether there was an alternative to modelling data where none exist (Murray et al.2001) and highlighting the uncertainty levels around the estimates (Murray et al.2000b), although even now, a decade later, these different views have not been resolved. Finally, to assess performance, it was necessary to estimate how much a health system could reasonably expect to achieve, given the myriad other influences on health. This estimate was derived from years of education, which, as Williams (2001) has noted, is an extreme over-simplification. This brief analysis begs the question of whether the report should have been published? Was it premature, given the scarcity of real data and the many unresolved methodological issues? Arguably it was not. Notwithstanding its many limitations, the report did manage to place assessment of health system performance firmly on the political and research agendas. It also provided much needed clarity on the elements of a health system. Thus, it clarified the concept of resource generation, emphasizing that it is not sufficient simply to provide more money for health care, but it is also necessary to think about how to supply those things that make it possible to deliver care, including trained staff, pharmaceuticals and appropriate facilities. It also introduced the concept of stewardship, a word that has defied translation into other languages but which captures a set of activities that involve the careful and responsible management of the health systems that have been entrusted to governments (Saltman and Ferroussier-Davis 2000). Crucially, however, it stimulated a major programme of research and analysis. A clear priority was to address the scandal of ignorance whereby far too many people are born, live and die without anyone recording their existence. Work by groups in the WHO itself, through the Health Metrics Network, the Institute for Health Metrics and Evaluation at the University of Washington at Seattle, and the In-Depth network of demographic surveillance, among others, has contributed to major improvements in availability of data, although there are still far too many gaps. Work has also proceeded on the task of estimating data where these gaps remain, as in the major effort to estimate maternal mortality in 181 countries (Hogan et al.2010), albeit that this is not an adequate substitute for having real data. Gaps in data on responsiveness of health systems were addressed by the World Health Survey, also involving methodological developments to overcome cultural differences in expectations. Thus, the use of ‘anchoring vignettes’ made it possible to standardize assessments of this element of health system performance (Salomon et al.2004). Other dimensions of performance have also been addressed by means of greatly increased use of existing survey data, such as the development of a measure of catastrophic expenditure on health care (Xu et al.2003). Notwithstanding these developments, problems remain. The World Health Survey promised to fill a major gap in our knowledge but it was effectively stillborn. The continued use of summary measures of overall health suffers from one major limitation, the problem of attribution. The 2000 World Health Report circumvented this problem by defining a health system as ‘all organizations, people and actions whose primary intent is to promote, restore or maintain health’. Yet, as several commentators have noted, while the health care system might plausibly be expected to advocate for health public policies, health is influenced by many determinants beyond the formal health care system, exemplified by the apparently very good performance in the 2000 report of countries known to have Mediterranean diets (McKee 2001). One approach, that has attracted growing interest, involves use of the concept of avoidable mortality, whereby deaths from causes that are preventable in the presence of timely and effective care are monitored (Nolte and McKee 2004). This approach has shown, for example, that while many industrialized countries have achieved substantial improvements in recent years, one, the USA, has been left behind (Nolte and McKee 2008). It is also possible to look at individual causes of avoidable death, such as diabetes, where there are also substantial differences in outcomes among industrialized countries (Nolte et al.2006). Data on cancer survival, once used mainly to track progress within countries, is also being used increasingly to compare outcomes among countries (Abdel-Rahman et al.2009), although here too there are a number of methodological caveats to be considered (Desai et al.2010). Efforts to use other measures, such as those based on data from health care facilities, are yielding useful experience, although also identifying challenges of comparability (Drosler et al.2009). The role of the 2000 report in stimulating the emergence of a community of researchers working on health systems performance should not be ignored. There are now a number of groups working to refine the available techniques. They include the teams in WHO and Seattle mentioned above but also others in OECD (Marshall et al.2006) and the European Observatory on Health Systems and Policies (Smith et al.2010). Yet perhaps the greatest achievement of the 2000 World Health Report was to place health systems performance on the political agenda. While some of the countries that fared poorly in the rankings simply ignored them, others commissioned research to discover reasons for their poor performance. They also asked questions of those providing health services. More generally, there are now a number of examples where seemingly poor performance compared with other countries has stimulated new policies, such as the once poor cancer survival in the United Kingdom (Abdel-Rahman et al.2009). It is now much more difficult for a politician to dismiss comparative data on performance; perhaps this is the report’s greatest legacy.