Background
Maternal and infant mortality are serious public health events and the majority of these are readily preventable. Classically, this has allowed them to be considered the best indicators of the standard of living and social well-being of a population [
1]. Their rates are presented as the annual number of female deaths from any cause related to or aggravated by pregnancy or its management during pregnancy and childbirth or within 42 days of termination of pregnancy, respectively, expressed per 100,000 live births; and the probability of an infant dying between birth and age of 1 year per 1000 live births, respectively [
2].
Infant mortality has two additional dimensions, the neonatal mortality rate, which is the probability of a child dying during the period from 0 to 27 days of life, expressed per 1000 live births, and the post neonatal period, in which a child dies within 28 days up to completing 1 year of age, expressed per 1000 live births. These dimensions are relevant, because the first month of life is the period in which the infant is most vulnerable and most susceptible to dying, from the point of view of health care organizations [
2,
3].
Over the years, the World Health Organization (WHO) and the United Nations (UN) have coordinated global efforts to reduce these rates, which have declined dramatically throughout the world. However, the desired goals of the Millennium Development Goals were not met, because despite the reduction of approximately 45% in maternal mortality and over 50% in neonatal and infant mortality, these events occurred in heterogeneous ways, with smaller or even lower reductions in the most vulnerable populations [
4].
To reduce these rates, the WHO recommends postnatal care in facilities for at least 24 h after birth, and postnatal contact within 24 h after birth for newborns born at home. Thus, interventions are focused on the delivery period and soon after birth, based on the cause of death, with some efforts to reduce neonatal deaths beyond the first week after birth [
5].
In recent decades, the struggle against maternal and infant mortality has intensified in Brazil, with public policies and greater allocation of resources to reduce it. With the aim of coordinating various sectors for the purpose of improving the quality of life of women and children, in 2004, Brazil established the “Maternal and Neonatal Mortality Reduction Pact”. The principles of this Pact are as follows: respect for the human rights of women and children; consideration of gender issues, ethnic and racial aspects, and social and regional inequalities; the political decision to invest in improving obstetric and neonatal care, including wide mobilization and participation of social managers and organizations [
6].
With these actions, it was possible to make significant progress in reducing infant mortality rate (IMR) to a rate of 14 deaths per thousand live births in 2013, since it anticipated Millennium Development Goal 4, which was 15.7 per thousand live births, in addition to representing a fall of 78% in this rate between 1990 and 2013 [
3]. In turn, the state of São Paulo recorded an IMR of 31.2 in 1990, which dropped to 11.66 per one thousand in 2013 and 11.4 in 2014, with a significant reduction since 2010 [
7]. However, when we analyzed the dimension of neonatal mortality, we found that the country maintained high levels, with a rate of 11.2 deaths per thousand live births in 2010. These mortality levels are below the levels the country is potentially capable of achieving, and can be considered reflections of the unfavorable conditions of life of population and health care, in addition to historical regional and socioeconomic inequalities [
6,
8].
When considering maternal mortality (MMR), despite the reductions observed, the results are less encouraging. The situation in Brazil has improved, as a decline in maternal mortality was shown from 141 per 100,000 live births in 1990 to 68 per 100,000 live births in 2010, but the goal of reducing ¾ of the target between 1990 and 2015 has not yet been reached [
9]. Between January and September 2011, the maternal mortality rate declined by 21% [
6,
8]. In the first decade of the twenty-first century, in the State of São Paulo, MMR was not expected to achieve the Millennium Development Goal 5: “Improve Maternal Health by presenting MMR equal to or lower than 35 deaths per 100,000 live births by 2015”. This was because only a slight drop was observed in this period, considering that in 2000, there were 275 maternal deaths (40.0) and in 2011, there were 249 deaths (40.8) [
10]. Considering the magnitude of this problem, little change was produced in maternal mortality relative to the prenatal period. Thus, there was a need for more knowledge about this relevant issue, with the aim of promoting the health of pregnant and postpartum women [
11].
Over the last decade, several studies have proven that both IMR and MMR were mightily related to contextual variables. The American study showed that improvements in primary care expressively influenced IMR and the number of underweight children in the US, thus demonstrating a negative association between the implementation of primary health care and reduction in infant mortality in the USA [
12]. However, few studies have discussed how political and economic contexts shape the effects of health and environment, so that a politically and economically unstable society would have difficulties with transferring health resources for effectively implementing care measures for the populations’ health, in spite of this society having a sufficient number of professionals and health facilities to meet these requirements [
13]. Therefore, research has the important task of reducing maternal mortality and its multi-causal factors around the world.
Although the family health care strategy (FHS) has a significant concentration of professionals and health services because of the latest measures implemented, it may still present the typical weaknesses of its immaturity. Therefore, it is necessary to carefully evaluate the different factors that may have affected maternal and infant mortality in the state, including the time since implementation of the FHS. The authors believe that this is the only way that will make it possible to discriminate the factors and their impacts without bias, and thus, to focus efforts on those that can effectively lead to improvement in these indicators.
Therefore, this study aimed to evaluate the relations between socioeconomic, demographic factors, the health care model and infant mortality (considering both the neonatal and post-neonatal dimensions) and maternal mortality in the state of São Paulo.
Discussion
For the analysis of maternal and infant mortality in the state of São Paulo, our study considered the social and macroeconomic context and the variables related to the health care model. The state of São Paulo is economically privileged; however, it has not reached the Millennium Development Goal related to maternal mortality. In contrast, it reached the infant mortality rate target before the agreed period. Thus, the importance of in-depth study of the factors that affect these indicators is evident. It is fundamental to include variables linked to the health care model, as the state of São Paulo implemented the FHS in a slower and more heterogeneous way than most of the other states in the country, because this state already had a structured network of health services in 1994 (when implementation of the FHS began). Politically and technically, it was believed that even with federal incentives, it was not worth changing the model of care. This view has changed over the years.
Shi et al. verified that the effect of income inequality on infant mortality disappeared when the statistical analysis led to adjustment of the model for areas where there was an increase in primary health care coverage, especially in the regions of greater social inequality [
12]. This emphasized the importance of an appropriate statistical model that took the model of care into account. In this study, we considered the variables related to the economic context and linked to the model of care, also including the time since implementation of the model (first and last implementation) to control the effect of the previously mentioned heterogeneity of FHS implementation in the state.
Optimizing infant health, by improving health monitoring, nutrition, vaccination schedules and living conditions, and implementing parental education programs to minimize unintentional injuries, may lead to reducing injuries to vulnerable children [
19]. In addition, better socioeconomic and demographic conditions, and the development of population health seemed to have caused the decrease in the maternal mortality among African descendants, where higher gross domestic product (GDP) was associated with lower MMR [
20]. While in 2014, Lourenço et al. [
21] found that as there was growth in the GDP from 1998 to 2008, infant mortality decreased in the state of São Paulo, Brazil.
In another study [
22], the WHO verified that almost all maternal deaths occurred in developing countries (99%). These countries have major public health problems. They struggle against poverty and suffer from limited access to health services, including antenatal care and childbirth. Optimizing child health by improving health monitoring, nutrition, vaccination schedules and living conditions, and implementing parental education programs to minimize unintentional injuries could reduce injuries in vulnerable children [
19].
In Africa, the results of Dersarkissian et al., in 2013 demonstrated that better socioeconomic and demographic conditions, as well as population health development, appeared to cause a decrease in African maternal mortality, where a higher gross domestic product (GDP) was associated the lower RMM [
20]. In Brazil, with the growth in GDP from 1998 to 2008, decreased infant mortality was found in the state of São Paulo, Brazil [
21].
Although limited, evidence on community-based primary health care (CBPHC) has been shown to be indispensable for improving maternal, newborn and child health, and to provide guidance for reducing the still prevalent indicators of preventable maternal and infant mortality [
23]. In Brazil, the state and national child nutrition programs adopted in recent decades have influenced the drop in infant and maternal mortality [
24]. Other studies have provided convincing evidence that the measures for reduction of income disparity produced improvements in the health of children and mothers in a short time interval [
25,
26]. Thus, the variable “model of care” always needs to be present in the studies on infant mortality, irrespective of the analysis made.
In 2014, Lourenço et al. [
21] affirmed that there was a decrease in infant mortality in the state of São Paulo from 1998 to 2008. However, the authors found no significant difference between 2004 and 2008, and no difference in the non-continuity of impact of the care model on the drop in IMR, indicating the need for further studies to better investigate the issue of infant mortality after 2008.
In addition to analyzing a period subsequent to the one studied by the authors, as advocated by the cited study, the present research verified that data such as considering the deployment time of the care model, not only clarified the importance of the FHS in infant mortality, but also of its continuity. Our study confirmed the importance of the care model (in this case, the FHS) in the control of infant mortality when we found the significant association of IMR with both the time of starting implementation and time of the last implementation, and even with the proportion of FHS deployed. Furthermore, this study reaffirmed the importance of performing this model throughout the maternal and child cycle, as we found that this effect occurred not only on the general IMRs but also on their neonatal and post-neonatal components.
Another important finding of this study was the association of both neonatal and post-neonatal mortality rates with birth rates. Mendes et al. [
27] in 2012, in a study conducted with the Brazilian population, affirmed that the drop in the birth rate, reduction in fertility and decrease in the population of children under one year old determined the greater care families took of their children and better health assistance provided by the Brazilian State. Moreover, according to the authors, pediatric care was very sensitive to social and epidemiological changes. This situation could also be verified in countries such as Morocco, where in spite of the actions and strategies adopted, persistent socioeconomic inequalities continued to hamper improvements in indicators of maternal and child health [
27].
The association of neonatal infant mortality with cesarean delivery supported the statement that the lowest perinatal mortality rates corresponded to those of countries that have cesarean rates below 10%, according to the WHO [
28], and this was related to the care model. However, according to Patah and Malik [
28], analyses of cesarean section rates should be done in the light of the health care model in force and social and cultural characteristics of a given society. The model of obstetric care defined, doctor-patient relationship, economic incentives and use of medical technology were of extreme importance in achieving the cesarean delivery. Considering the possibility of a better doctor-patient relationship in the FHS model, based on continuity of care, it is expected that pregnant women assisted by the FHS will receive information and empowerment to at least take part in the decision about the type of delivery. This should decrease the chances of cesarean section if the patient has been properly informed about the risks and benefits of all types of delivery. In addition, the association here verified also reflected certain data related to the socioeconomic factor.
In regions where there was greater coverage by health plans, the rates of cesarean section were also higher [
28]. This was because private physicians were usually paid by production and the cesarean section was better remunerated than vaginal birth and demanded less time to be performed, in addition to the possibility of performing the section by prior appointment. In 2013, the state of São Paulo had the largest coverage of private health insurance in the country (44.41%) [
10]. Despite this, as far back as 2004, Dias and Deslandes [
28] stated that even in public maternities, cesarean section rates were still higher than expected, because surgical indications were also governed by issues related to medical training and to cultural trends of assistance.
There are several complex issues related to medical training and culture, model of obstetric care and working processes that need to be reviewed so that these factors fit the national cesarean section rate, especially in the state of São Paulo, because other complex questions also affected the neonatal infant mortality rates.
Much has been done in the country, such as the greater appreciation of Primary Health Care by implementation and expansion of the FHS, insertion of women in the labor market, increased education and reduction in the inequality of income, but the reduction in MMR has taken place at a slower pace than was expected. In the state of São Paulo (HDI = 0.783) - the richest in the nation [
10] - there was an even more intriguing situation because the MMR remained stable from 2000 to 2011, although the rate was much lower than the national rate in 2000. However, there was an approximation to the national rate in São Paulo, due to the continuous drop in the national rate and stagnation of the state rate.
Ruiz et al. [
1] recommend that the HDI adjusted for inequality was the best predictor for IMR and MMR. Income inequalities in São Paulo and in Brazil, as measured by the Gini index, were 0.472 and 0.501, respectively. However, despite the numerical difference, evolution of these indices occurred at the same rate in São Paulo and in the country from 2004 to 2013. While Brazil went from a Gini index of 0.555 in 2004 to 0.501 in 2013, São Paulo went from 0.5239 to the present 0.472. This pointed out the fact that the better income conditions of classes C, D and E in Brazil in the last two decades, added to the fact that there were expanded and improved primary care networks of health services throughout the country may have contributed to the significant improvement in MMR. In São Paulo, where these improvements in income also appeared to have occurred, there was heterogeneity and slowness in the implementation of the FHS, especially in the first decade of the 2000s. Therefore, this factor seemed to be decisive in explaining the slower drop in the indicator in a state with socioeconomic conditions such as those of São Paulo. Health surveillance – pillar of the FHS – was the feature that enabled the health teams to “capture” the pregnant women in the territory, who were still unattended, thus providing them with access to prenatal care the timeliest possible way, thereby overcoming an important factor of impact on maternal and child mortality - the lack of access or unqualified access (with insufficient visits) to prenatal care.
As regards the care model, needed reduction in cesarean section rates in Brazil, and in this case, in the state of São Paulo, the need to review the medical education in Brazil is highlighted. This is still rooted in the biomedical hegemonic model, hospitalization and dependence on hard technologies. The family health model requires practices that focus on interdisciplinary, surveillance and teamwork actions. The reduction in cesarean sections is a topic that needs to undergo reformulation of the curricula of medical schools to emphasize the humanization of childbirth, importance of pregnant women’s autonomy for making the decision about the type of delivery, and especially empowerment of the physician to make a decision jointly with the pregnant woman, so that their decisions take precedence over financial or operational and administrative questions.
Limitations of the study and suggestions for future researches
The major limitation of this study was that the relationship between the factor of exposure and the event might not occur at an individual level because it was an ecological study. However, the analysis made it possible to identify the factors that deserved a more detailed investigation by means of other study designs. The fact that this was an ecological study with cross-sectional analysis was an outstanding limitation, as it hindered the establishment of causal links, however, there was a longitudinal linkage by previous data collection, such as the variable “time of implementation of the FHS” (since 1998).
Moreover, we suggest the continuity of this study through longitudinal researches to verify maternal health care from the perspective of cause and effect. However, there are two other suggestions: the first is focus on the group denominated “
Near Miss”- composed of women who present potentially lethal complications during pregnancy, which translates into a group with better and more frequent sources of information in cases of maternal death [
29,
30]. The study of this group should allow determination of the variables that are capable of better elucidating the difficulties in reducing this indicator. The second suggestion would be to check the impact of the “Stork Network” in maternal and infant indicators, since the purpose of this important healthcare network implemented by the Ministry of Health was to structure and organize the maternal and infant health care in the country. However, as it was implemented in 2011 and our study used data from 2013, this network was not analyzed here, as we understood that conclusions could be unfeasible, hasty or biased.
Another limitation requiring consideration is related to the Information Systems available in Brazil, which may contain database fragmentations, under-registration or even methodological inconsistencies. In the poorer municipalities, failures in the recording systems are known to be even greater [
31,
32]. The major challenge concerns the inadequate completion of the Declarations of Live Births and Deaths (DN and DO), although there have been undeniable improvements in registration systems in the first decade of the century and modification of the declaration models in 2011. This has led to more efficient registration of variables such as, for example, prenatal care [
33]. However, in relation to SIM, the study by Maia et al. found high percentages of incompleteness, and the only variable component of the statement that obtained “optimal” classification was the gender of the child [
34]. Other studies at national level have also highlighted the high percentage of incompleteness in SIM as regards the duration gestation, an important predictor of infant death [
34,
35]. These deficiencies in filling out important variables have led to limitations in the potential use of the system for epidemiological studies [
36]. However, although these limitations exist, it is important to note their frequency is low and they would certainly not change the conclusions of this study.