Background
Preterm birth, which is defined as birth before 37 weeks of gestation, is a global phenomenon of increasing prevalence [
1], which has also been observed in earlier Brazilian studies, including the Pelotas cohort wherein preterm births increased from 5.8% (1982) to 13.8% (2015) [
2]. Global trends indicate that estimated prevalence of preterm births varies from 9.6% to 11.1%, and low- and middle-income countries are responsible for most of the burden [
3,
4]. An increase in the preterm birth rates is attributable to both, an increase in medical indications and in artificial gestations [
5], which have enhanced the rate of elective preterm births. Moreover, improved national information systems and advances in perinatal support are important factors for this increase and the higher life expectancy of children who are born preterm [
1].
The mechanisms that lead to birth before 37 weeks of gestation are unclear and constitute an expanding field; however, three main factors for spontaneous preterm birth have been proposed: social stress and race; infection and inflammation; and genetics [
6]. The consequences of preterm birth include higher perinatal morbimortality [
5,
7] and an impact on child development and health as well as well-being in adulthood [
3].
Most of the population-based studies that investigated the related factors and the incidence of preterm birth have been conducted in developed countries. In less developed countries, the relevant research tends to be observational and mostly single-centre studies, due to which the results are not representative of the general population [
3].
Previous studies in the United States found that area poverty and deprivation, assessed by schooling, housing, occupation, race, and income indicators, are related to a higher risk of preterm birth [
8,
9]. Furthermore, a meta-analysis that included 28 studies showed that lower-income neighbourhoods were associated with a higher risk of preterm birth [
10]. A longitudinal study in Spain identified a higher risk of spontaneous preterm birth among pregnant women from countries that had a low or medium Human Development Index (HDI) [
11].
Nonetheless, a Dutch study showed contradictory results: non-Western pregnant women who lived in neighbourhoods with a lower social index were not at a higher risk for adverse perinatal outcomes than those who lived in higher index locations [
12]. Moreover, after adjusting for multiple comparisons, a study that was conducted in Hong Kong found no association between neighbourhood income, inequality, or the Gini Index and preterm birth [
13].
This study aimed to contribute to this literature gap, identifying whether less-developed and equal areas of Brazil are associated with a higher incidence of preterm birth, specifically spontaneous and elective preterm births. The different socioeconomic factors associated with spontaneous and elective preterm births are not commonly explored features in contemporaneous literature.
Methods
This retrospective observational study used nationwide data that were obtained from national database of live births, the Sistema de Informações sobre Nascidos Vivos (SINASC), and the Brazilian census conducted by the Brazilian Institute of Geography and Statistics (IBGE).
Preterm birth rates from 2014 to 2019 were analysed by the Brazilian Federal Unit (FU) and by year (Additional File
1). A pre-analysis showed that data from 2014 were more accurate due to recent improvements in preterm birth measures and notifications [
14‐
16] and 2019 was the last year which had consolidated data; therefore, the interval from 2014 to 2019 were considered for analysis in this study and data from 2019 were selected for further analysis. Data were extracted from the Data Science Applied to Health Platform (PCDaS) [
17] of the Oswaldo Cruz Foundation (Fiocruz
).
Census data were extracted from the Pesquisa Nacional por Amostra de Domicílio (PNAD) of 2017, and included HDI and its sub-components – life expectancy, education, and gross national income (GNI) per capita; the Gini Index; illiteracy; access to water; access to sanitation; and life expectancy of Brazilian FUs. Preliminary analysis showed no significant variation in the above-mentioned indices throughout the study period; accordingly, the most recent data from 2017 for all indices, except for life expectancy (for which the most recent data was from 2018), were used. Socioeconomic metrics were extracted from the Atlas of Human Development in Brazil (
Atlas Brasil) [
18] and from the Health System Performance Assessment Methodology (PROADESS) platform. The SINASC and IBGE census records contain publicly available anonymised data. Therefore, with regard to the Brazilian ethics regulatory requirements, no prior informed consent or ethical approval by the institutional review board were required [
19].
Spontaneous delivery comprised all forms of non-induced labour except for elective caesarean birth, remaining only cases with spontaneous onset of labour. Elective delivery was defined as provider-initiated, which included induction of labour and elective caesarean birth (either because of maternal or foetal indications, or other non-medical reasons) [
20].
The preterm birth rate was calculated as:
$$\mathrm{Preterm}\;\mathrm{birth}\;\mathrm{rate}\;=\;\mathrm{Number}\;\mathrm{of}\;\mathrm{births}\;\mathrm{from}\;22\;\mathrm{to}\;37\;\mathrm{weeks}\;\mathrm{of}\;\mathrm{gestation}\;/\;(\mathrm{total}\;\mathrm{number}\;\mathrm{of}\;\mathrm{births}\;-\;\mathrm{number}\;\mathrm{of}\;\mathrm{births}\;\mathrm{missing}\;\mathrm{for}\;\mathrm{gestational}\;\mathrm{age})$$
The spontaneous preterm birth rate was calculated as:
$$\mathrm{Spontaneous}\;\mathrm{preterm}\;\mathrm{birth}\;\mathrm{rate}\;=\;\mathrm{Number}\;\mathrm{of}\;\mathrm{births}\;\mathrm{with}\;\mathrm{spontaneous}\;\mathrm{delivery}\;\mathrm{from}\;22\;\mathrm{to}\;37\;\mathrm{weeks}\;\mathrm{of}\;\mathrm{gestation}\;/\;(\mathrm{total}\;\mathrm{number}\;\mathrm{of}\;\mathrm{births}\;-\;\mathrm{number}\;\mathrm{of}\;\mathrm{births}\;\mathrm{missing}\;\mathrm{for}\;\mathrm{gestational}\;\mathrm{age})$$
Similarly, the elective preterm birth rate was calculated as:
$$\mathrm{Elective}\;\mathrm{preterm}\;\mathrm{birth}\;\mathrm{rate}\;=\;\mathrm{Number}\;\mathrm{of}\;\mathrm{births}\;\mathrm{via}\;\mathrm{elective}\;\mathrm{delivery}\;\mathrm{from}\;22\;\mathrm{to}\;37\;\mathrm{weeks}\;\mathrm{of}\;\mathrm{gestation}\;/\;(\mathrm{total}\;\mathrm{number}\;\mathrm{of}\;\mathrm{births}\;-\;\mathrm{number}\;\mathrm{of}\;\mathrm{births}\;\mathrm{missing}\;\mathrm{for}\;\mathrm{gestational}\;\mathrm{age})$$
The fraction of spontaneous preterm births was defined as the number of births via spontaneous preterm delivery from 22 to 37 weeks of gestation divided by the total number of births from 22 to 37 weeks of gestation, and the fraction of elective preterm births was calculated as the number of births with elective delivery from 22 to 37 weeks of gestation divided by the total number of births from 22 to 37 weeks of gestation.
The data were summarised as absolute and relative (%) frequencies. Spearman’s correlation was used to determine the correlation between socioeconomic indicators and the preterm birth rate. Multiple analyses were performed using a beta regression model with a logit link function [
21], because the preterm birth rate varies within the interval (0,1). As we have only 27 subjects (27 FUs) for use in the multiple regression model, we performed a pre-selection analysis, which ascertained whether the Spearman’s correlation between two socioeconomic variables was greater than 0.8. The factors with the highest correlation with the preterm birth rate were selected as covariates for inclusion in the beta regression model for preterm birth rate.
The likelihood ratio test for heteroskedasticity was performed to assess which of the following regression models was more suitable: fixed or variable dispersion beta regression (where Φ is the dispersion parameter) [
22]. For preterm birth, specific regressions were performed for ascertaining the spontaneous and elective preterm birth fractions. Akaike Criteria [
23] were used to determine the best model and diagnostic tools in beta regression were performed to evaluate model fit and assumptions [
22]. The factors importance was presented in terms of permutation feature importance graph by considering the gain in terms of Mean Square Error (MSE) [
24]. The significance level was established at 5%. All analyses were performed using R (
https://www.r-project.org, version 4.0.3).
Discussion
To our knowledge, this is the most comprehensive study conducted in developing countries regarding socioeconomic influence on preterm birth rate. We observed that areas with lower levels of development in Brazil were associated with higher spontaneous preterm birth fraction, whereas areas with higher development status were associated with higher fraction of elective preterm birth.
In our study, we found that, in Brazil in 2019, the mean prevalence of preterm births was 11.03% (range, 9.03%–21.38%), which aligns with previous estimates from other studies wherein the prevalence varied from 10.2% to 15% [
25,
26]. The last national multicentre study, conducted from April 2011 to July 2012, [
27] found a mean prevalence of 12.3%, which is compatible with other recent studies in a FU capital city [
28] and with the results of our data analysis. The higher rate of preterm birth in Amapá constitutes an outlier and should be further analysed by temporally comparing the series and consistency of the public register in future years.
With regard to the national fractions of spontaneous and elective preterm births in 2019, 58% were spontaneous, with higher rates in the northern region and lower rates in the south and southeast regions. This finding represents a reduction in the fraction of spontaneous preterm births that infers an increment in the fraction of elective preterm births, compared with the results of a 2014 study, which found that 64.6% of preterm births were spontaneous [
27]. Higher rates of spontaneous preterm birth in the north, northeast, and mid-western regions were observed previously [
29]. Elevated rates of spontaneous preterm births are associated with regions that have a lower socioeconomic status [
30], and a possible proximal explanation may rely on differences in prenatal care coverage [
29,
31], which we were unable to verify in the present study.
Considering the total rate of preterm birth, we found a direct relationship between the HDI, the Gini Index, and life expectancy, but discerned an inverse relationship between illiteracy, access to water, and sanitation. Our results present a similar paradox to that observed in other Brazilian studies [
28,
32], wherein low birth weight was higher in more developed areas. This could be explained by higher medical assistance and better registration in areas with a higher HDI. Positive association with Gini Index and total preterm births could be explained as a higher burden of spontaneous preterm birth among total births, as well as because of Brazilian elevated indexes of inequality in richer areas.
When analysing spontaneous preterm births, we found a direct relationship between illiteracy and the Gini Index and an inverse relationship between HDI, life expectancy, access to water, and sanitation. Our results indicate that areas with lower levels of education and less equality in income distribution, which are risk factors otherwise proven to be relevant, have a higher fraction of spontaneous preterm birth [
31,
33]. Higher HDI and life expectancy, and greater access to water and sanitation all represent higher socioeconomic status that, therefore, predicts a lower spontaneous preterm birth rate, as shown by our data, [
30].
Finally, elective preterm birth presented the converse associations of the spontaneous preterm birth analysis. Higher elective preterm birth rates are associated with advances in obstetric procedures and early diagnosis of maternofoetal conditions, all of which occur more often in richer and more developed locations [
34]. Increased elective preterm birth rates are commonly accompanied by lower infant mortality and stillbirth rates [
34], which indicates better perinatal support. Another important factor is to ascertain whether richer areas may be overproducing preterm births, which could be due to increased access to medical assistance and more incisive surveillance of maternal-foetal conditions.
The main strength of our study was the use of national data, which provided a broad and comprehensive outlook of preterm birth in Brazil. Another unique aspect was the approach adopted to determine both spontaneous and elective fractions of preterm births, which has rarely been explored in Brazilian studies. The use of state-of-the-art statistical methods provides an enriched analysis.
The study’s limitations emerge primarily from the use of public national systems of information. Despite recent enhancements in the quality of SINASC data, underreporting remains an issue [
35], especially in less developed areas of the country. Inappropriate records can lead to misleading correlations. Moreover, the lack of more recent census information was detrimental to the accuracy of the analysis, as births in 2019 were compared to socioeconomic indicators, mainly from 2017. Nonetheless, we believe that the impact was minor due to the stability of these metrics over short periods, although this aspect surely limited the interpretation of the results.
Conclusions
Our study showed that higher socioeconomic status was associated with a higher fraction of elective preterm birth rate and, complementarily, lower spontaneous preterm birth. Preterm birth rate substantially differs depending on the development status of the FU, as well as its distribution in elective and spontaneous births.
The study provides unique and comprehensive continent-sized and population-based data, which can potentially be used as a powerful tool to orient policy decisions aimed at reducing perinatal complications. Orienting our view on preterm birth by two different entities – spontaneous and elective preterm births – allows targeted interventions according to territory socioeconomic parameters aimed at reducing the rate of preterm births and ensuring better outcomes in the morbimortality of infants and mothers.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.