Many countries in Latin America have undergone substantial reforms of their health care systems in the last twenty years. Measuring their impact on health outcomes is crucial to understand how effective they have been in achieving their stated goals [29
] of promoting better access to health care services, financial security and reducing health inequities. To this date, most evaluations [23
] of those reforms have been limited to single country analyses based on national cross-sectional data. The disadvantage of such methodology is that it does not adequately control for existing underlying time trends present in the variables of interest at the time the reforms were enacted. Furthermore, comparisons between countries are difficult to assess based on those individual studies since they often entail different variable definitions, different time periods and survey methods.
The present work thus attempted to redress those issues by utilizing internationally standardized time series data for Brazil and Colombia, covering the period from 1960 to 2005. Another innovation introduced by this study was that it is the first to compare the achievements in health functioning for Brazil and Colombia within the context of the health capability paradigm
]. As a result, not only did it analyze their individual performances in the four health outcomes of interest (crude mortality rate, infant mortality rate, under-five mortality rate
and life expectancy
), but also assessed their success in reducing the shortfall inequalities in those variables when measured against the corresponding optimal values in Latin America and the Caribbean.
The results of the analyses of the baseline values of each country showed that while both of them succeeded in improving those indicators throughout the time period of the analysis, the impact of their health care reforms was much less tangible. In fact, in most cases they were not able to alter the underlying time trend already in place when the reforms were enacted. However, when they did influence the outcomes, the outline was inevitably negative. In other words, the years after the reform often saw the deceleration of the pace of improvements in both countries for all the variables analyzed. This effect was even more pronounced in Colombia, where some of the previous gains in the under-five mortality rate and the crude mortality rate were reversed.
The subsequent analyses of the shortfall inequalities further aggravated the perception that the health care reforms of Brazil and Colombia did not contribute to the reduction of heath inequities. When controlling for the underlying time trend, neither reform produced a statistically significant positive contribution to any of the variables studied. On the contrary, their respective impacts were often counterproductive. Consistent with the above results, Colombia fared once again worse than Brazil. In fact, both in life expectancy and crude mortality rate the average measured effect of the reform years overturned the historical time trend and effectively increased shortfall inequalities among those variables. Also worth noticing is the fact that GDP per capita, used as a control in the second model of the multivariate linear regressions, only had a statistically significant impact in two of them, namely: shortfall inequality of crude mortality rate in Brazil (p = 0.0242) and shortfall inequality of life expectancy in Colombia (p = 0.0262). When it did so, its influence was to increase the inequities.
Given the public praise received by both reforms from domestic [22
] and international [8
] observers in recent years, the findings of this study seem counterintuitive. How can this be true? First, most previous studies did not explicitly control for the effect of the historical time trend on results measured after the reform. Second, when they did control, it was within a differences-in-differences design for pooled cross-sectional data, often with just one pre-reform time period and one post-reform time period. Third, nearly all of them used national data, some of which might not have been adjusted to international standards, making cross-country comparisons much more difficult. Fourth, none of them measured the reforms of Brazil and Colombia against an independent third-party reference value in order to assess the effective reduction of the gap between their current achievements and the optimal international average (i.e. shortfall inequality) for the region.
Still, the present work has some limitations that must be acknowledged. First, the scarcity of internationally standardized data for health outcomes, especially for time points before 1990, severely restricted the diversity of outcomes that could be analyzed, thus forcefully limiting the scope of this study to the four health outcomes mentioned earlier. Second, because of such scarcity, a 5-year interval between data points had to be observed rather than an annual or semester basis, thereby increasing the influence of each individual data point in the overall fit of the regressions. Third, because of gaps in the dataset a few data points had to be estimated from 5-year averages, increasing standard errors and reducing the likelihood of statistically significant results. Fourth, the cut point of this study was set for 2005, when the Brazilian health care reform had been established for just over 15 years while the Colombian reform had little more than 10 years, both of which might be too short to comprehensively evaluate their long-term effects. Fifth, despite the efforts of ECLAC and the World Bank to standardize the data across countries, differences in methods and definitions might still account for some of the variability observed in the results of this study. Sixth, the attempt to control for the underlying historical trends assumes that such trends would remain constant in the years after the reforms were enacted, which does not account for countries' susceptibility to unforeseen economic shocks and social upheavals. Seventh, in the absence of health care reforms, it is possible that historical trends might not have remained constant throughout the time-series, given the potential plateau effect related to the ability to maintain constant changes as levels of performance rise, such as when outcomes improve.
In spite of such limitations, the high adjusted-R2 of most of the models and their consistent pattern across the four studied variables suggest that the impact of the health care reforms in Brazil and Colombia was not nearly as positive as expected. Why? Some of the answers might lie outside the realm of the health care system while others could be nested in the heart of the design of each health care reform.
Within the realm of health care reforms, some features are likely culprits. For instance, in both Brazil and Colombia insufficient funding is likely to have slowed the implementation of new services that could have helped alleviated suffering populations. By extension, the overt emphasis on decentralization might have led to the disruption of previously functional vertical programs aimed at child survival, which in turn might have adversely effected the infant and under-five mortality rates. In the case of Colombia, the explicit tiering of the social insurance model into a contributory regime and a subsidized regime with distinct packages of benefits might have crystallized health inequities due to uneven access to care. Furthermore, such an arrangement is inherently unfair according to the health capability paradigm [3
], as it denies equal access to high-quality care, an essential requirement in order to secure the achievement of central health capabilities. Moreover, the latest ruling of the Colombian Constitutional Court stroke down the two-tiered system, considering it unconstitutional and demanding a complete overhaul by 2010. On the other hand, although the Brazilian public health care system (SUS) is nominally universal and every citizen is entitled to comprehensive care, high regional inequities in access remain [36
], thereby perpetuating long-lasting health disparities, which in turn hinder progress in the outcomes analyzed by this study. For instance, a recent study [37
] presents a thorough analysis of sub-national inequalities in Brazil for under-five mortality rates and neonatal mortality rates which demonstrates that, in general, the Brazilian health care reform has been much more effective in improving rates in the better off than the worst off municipalities of the country. In many cases, this has led to widening poor/rich gaps.
Outside the health sector, both countries have endured imposing challenges. In the economic arena, the currency crises in Russia, Mexico and Southeast Asia that happened in the 1990's severely battered the overall economic stability of both countries. In this regard, Colombia seems to have suffered the most, since its economy is more dependent on the external market than the Brazilian economy. The graphs seem to corroborate this hypothesis, as a drastic slowdown in the rate of reduction of infant and under-five mortality rate starting at 1990 is unmistakable for Colombia but not as meaningful for Brazil. In addition, Colombia has been embattled in a fierce fight against domestic insurgent groups, particularly after 1990, with likely adverse effects in its crude mortality rate and life expectancy. Likewise, Brazil has also had to grasp with growing urban violence since 1995, possibly resulting in a slowdown of any progress in the reduction of its crude mortality rate. Nevertheless, no significant impact of this problem has yet been noticed in Brazil's life expectancy, perhaps because its effect might take longer to accrue. Another relevant aspect of the social and political scenario of both countries that is likely to be thwarting progress in health capabilities is the high level of income inequalities, as described by Gini coefficients of 0.569 for Brazil (2004) and 0.563 for Colombia (2003).