To our knowledge, this is the first study in Colombia and Latin America that has addressed the question about the contribution of the PHC strategy to reducing health inequalities using both the concentration index and decomposition analysis. Changes in the values of the concentration indices indicates that child health inequalities were reduced in 2007, the period after the implementation of the Home Health program in Bogotá. Results of decomposition analysis allowed us to establish the contribution of the PHC in the reduction of the inequalities, which range from 7% in acute malnutrition to 24% in IMR. The effect of the PHC on reducing inequality remained significant even in the presence of the other examined variables. Results of both the concentration indices and the decomposition analysis suggest that the increase of the Home Health coverage (through the expansion of the number of health personnel per population and interventions of the Home Health program in low-QLI localities) might have a positive effect on reducing disparities in the four indicators studied.
Our results are also consistent with evidence from other contexts where a positive impact of PHC has been reported on improving equity, mainly with regard to the reduction of disparities in access and health outcomes when analyzed by socioeconomic status, ethnicity and geographical location [19, 20, 31–41].
Although the experiences in the Latin American (LA) context vary regarding the implementation of PHC, studies have shown that the strategy can contribute effectively to reducing gaps in access to and the use of health services associated with socioeconomic differences. This could be partly explained because the implementation of PHC in most LA countries usually began by prioritizing the most disadvantaged groups [20]. In addition to reducing disparities in access, PHC has also shown success in reducing disparities in child deaths. For example, research from Mexico [31] found that certain characteristics of the primary care delivery services (coordination, longitudinality and comprehensiveness) had an important effect on reducing the probability of infant death in socially disadvantaged areas. A study from Bolivia [32] found that a PHC approach focused on solving community needs and the promotion of social participation in socially disadvantaged areas reduced under-5 mortality rates more than in adjacent areas as well as the entire country. Another analysis of nine LA countries, which analyzed the effects of the economic crisis on the trends in infant mortality rates, suggested that those countries where IMR had declined and inequalities had not increased (Chile, Cuba and Costa Rica) were those where access to primary care services had also increased [33]. Specific analyses of the Costa Rican health system reforms have confirmed the results mentioned above. In the case of Costa Rica, the interventions included the expansion of the number of primary care facilities and the creation of basic health care teams assigned to a number of families. These interventions have been associated with increases in life expectancy, decreases in infant mortality rates and a reduction in inequalities in access [34]. For its part, Brazil has provided some evidence suggesting that the “Family Health Program” expansion in the north and north-east regions of the country may have contributed to reducing inter-regional inequalities in infant mortality [35].
Also, comparative country studies have shown the great potential of primary health care to reduce disparities in health outcomes associated with socioeconomic status. Thus, an analysis found that 90% of child deaths were concentrated in 42 countries, and 63% of these deaths could have been prevented by PHC interventions such as comprehensive care for diarrhea, pneumonia, measles, malaria, HIV/AIDS, preterm delivery, neonatal tetanus and neonatal sepsis [19, 36]. Other examples of improvements in health equity in developed countries have shown how PHC services have been associated with reductions in socioeconomic inequalities related to ethnicity and geographical location in health outcomes (child mortality and all-cause mortality) and self-perceived health status [37–41]. A literature review, including some studies from the United States of America, highlighted that better primary care development (measured by the number of primary care physicians assigned to the population) is associated with relatively greater effects on health status in socially disadvantaged areas (measured by high levels of income inequality) [19]. This review concluded that areas with high income inequality where PHC was better developed had lower infant mortality rates and better levels of self-perceived health status than areas of high income inequality with less PHC development. In addition, the adverse impact of income inequality on all-cause mortality was significantly reduced by strengthening primary health care interventions [19].
It is important to mention that, although our findings are consistent with other investigations and that the effect of PHC on equity was positive, the magnitude of its contribution to reducing disparities was relatively small for all indicators, especially with regard to acute malnutrition. A possible explanation for the small contribution might be that the indicators analyzed were mostly evenly distributed in 2007 (their concentration index had values very close to zero). It could also be because only data for the third year after the implementation of PHC were included in the present analysis, and a longer period of time would have been required to demonstrate a greater effect.
On the other hand, despite the overall sustained economic growth and poverty declines in Bogotá, inequalities in living conditions at locality levels have not changed substantially. That is the case in some localities where the PHC strategy has been better developed and where simultaneously levels of poverty have increased and coverage of health insurance has lowered [13, 42, 43]. Consequently, the pace of expansion of the PHC strategy might not have been sufficient to offset the increased vulnerability faced in some localities.
In addition to the previous argument, it is important to highlight that a wide range of social interventions that could affect child health outcomes, such as programs that provide economic and nutritional subsidies, have been implemented simultaneously with the PHC strategy. These interventions could make a greater contribution to reducing disparities, especially in acute malnutrition and child deaths, reducing the equity effect attributable to PHC intervention.
Likewise, it is known that the presence of basic health equipment and the increase in the number of professionals (variables included in the PHCI) do not necessarily ensure better access, use and quality of health services [19]. This is particularly relevant when other economic and administrative barriers (e.g. co-payments, fragmentation in the procurement of services, excessive paperwork requirements to access services and delayed care) persist in the Colombian health system [12–14, 21, 22], preventing the potential of the PHC strategy to impact on equity.
Finally, some other weaknesses that could also reduce the potential of PHC to affect inequalities include the persistent difficulties in linking all health system stakeholders. PHC has not been able to influence an adequate number of insurers and private providers to improve coordination. Community participation is still shaped according to the rules of the institutions and the rationality of the market, and intersectoral action has not been extended or deepened adequately [14, 15].
Study limitations
Our results are subject to the usual cautions of interpretation of cross-sectional results and the limitations of ecological analysis, which do not provide conclusive evidence of causality. The unavailability of information on a disaggregated level lower than localities (e.g. micro-territories, families or individuals) did not permit us to determine with certainty whether the reductions in disparities in the health outcomes were in favor of the vulnerable population reached by the Home Health program in the locality or if those were an average reduction.
Likewise, the few sources of information available to gather data on other variables recognized as determinants of inequalities in the health outcomes analyzed could be the reasons for residual factors contributing the greatest proportion of the inequalities. The complexity and influence of many social determinants on the health outcomes studied merit further analysis [22, 26].
Also, the analysis of the PHC contribution to reducing disparities requires further research, including additional variables that allow a better understanding of the effectiveness of its component health interventions (e.g. health education at facilities, home visits by community health workers, reference to social services) to discover which PHC interventions are contributing to improvements in health equity.
On the other hand, the periodicity of the socioeconomic information used in this analysis (which is collected every four years) reduced the possibility of including more observation periods, and this could have limited a better appreciation of the possible effect of PHC in reducing disparities. Indeed, measuring the implementation of the strategy only three years after its implementation might not be sufficient to measure its true impact on equity.
The QLI reflects a particular definition of living standards that is not necessarily equivalent to wealth or income levels, which are the variables frequently used to rank population in this kind of analysis. The QLI also summarizes many variables grouped in different dimensions. Different rankings of localities might be obtained if other indicators, besides QLI, were used.
One recognized limitation of the decomposition analysis is that the method relies on linear models [30]. In our case, the model used was a non-linear one due to the nature of the outcomes analyzed. The limitations related to the use of non-linear models, grouped data and a small number of explanatory variables could result in approximations to partial effects of these determinants on health outcomes.