INTRODUCTION

Exclusive breastfeeding is understood as feeding an infant breast milk only, either directly from a breast or extracted from the mother or a donor, and no other fluid or solid, except for vitamin, mineral, or medical drops or syrups¹. The actions for promoting, protecting, and supporting breastfeeding are effective and inexpensive strategies against childhood morbidity and mortality². The World Health Organization (WHO)³ recommends exclusive breastfeeding for six months and non-exclusive breastfeeding for at least two years, and the Brazilian Ministry of Health (MH)⁴ has supported the recommendation since 2001. Breastfeeding is closely related to the child's good nutritional and health statuses, and to greater resistance against infectious diseases, especially gastrointestinal infections.

Breastfeeding may reduce the preventable death rates of children less than five years of age by as much as 13% worldwide⁵. Although childhood mortality is decreasing in Brazil⁶, reducing preventable deaths is a permanent challenge, and the rates represent a public health problem that affects Brazilian regions differently⁷. In the North and Northeast regions, these rates are still high, so in 2009 the Pact for Reducing Childhood Mortality was signed in the Northeast and Legal Amazon to reduce these regional inequalities. The pact has six cores, one of them being prenatal, delivery, and newborn care qualification⁸.

Some Brazilian studies conducted at the end of the 20^th century covering a period of 25 years already detected a gradual improvement in the breastfeeding practices of the children less than one year of age, but the breastfeeding indicators varied significantly throughout the different geographic regions of the country^{9 , 10}. Brazil has a high percentage of mothers who start breastfeeding their children (95.0%), but exclusive breastfeeding does not reach the ideal duration, being much smaller than the recommended six months. The 2006 Pesquisa Nacional de Demo-grafia e Saúde da Criança e da Mulher (PNDS, National Child's and Mother's Demographic and Health Survey)¹¹ found an exclusive breastfeeding rate of 38.6% in children less than six months of age, with the Northeast region having the lowest median: 1.1 months. The II Breastfeeding Prevalence Survey¹² conducted in 2008 in Brazilian capitals and Distrito Federal found an exclusive breastfeeding rate of 41.0% in children aged less than six months, with the highest and lowest prevalences occurring in the North (45.9%) and Northeast (37.0%) regions, respectively.

Given this landscape, investigation of the possible factors associated with exclusive breastfeeding in children aged less than six months is one of the requirements for promoting exclusive breastfeeding policies and programs, and identifying the occasions that most contribute to early weaning. The present study assesses this aspect in municipalities of the Legal Amazon and Northeast regions in 2010 using data from the survey "Neonatal Call: assessment of prenatal care and care of infants aged less than one year in the North and Northeast regions".

METHODS

This study is based on secondary data from the "Neonatal Call: assessment of prenatal care and care of infants aged less than one year in the North and Northeast regions" conducted in 2010 and approved by the Research Ethics Committee of Escola Nacional de Saúde Pública Sérgio Arouca/Fundação Oswaldo Cruz (Ensp/Fio Cruz). The cross-sectional study included mothers of children aged less than one year who received the multiple vaccines in 252 of the 256 primary municipalities of the Pact for Reducing Childhood Mortality⁸. The municipalities are located in nine states of the Legal Amazon (Acre, Amapá, Amazonas, Ma-ranhão, Mato Grosso, Pará, Roraima, Rondônia, and Tocantins) and in eight states of the Northeast region (Alagoas, Bahia, Ceará, Paraíba, Pernam-buco, Piauí, Rio Grande do Norte, and Sergipe).

The Neonatal Call used probabilistic sampling to select the children vaccinated in each planned domain: the capital of each state and all other municipalities of the state. The minimum sample size of each domain, 750 mother/child dyads, was estimated considering a delivery complication of 22.0%¹³, an error <3.5%, a confidence coefficient >95.0%, and a design correlation factor (deff) <1.5, totaling 23,399 interviews. The mother/child dyads were selected by two-stage cluster sampling¹⁴ and probability proportional to cluster size. The first stage randomly selected the main vaccination stands in each municipality; the second stage consisted of a systematic selection of mother/child dyads at each vaccination stand, according to the study inclusion criteria. Mothers of children aged more than one year, living in other municipalities, and whose children were twins and/or adopted were excluded. A total of 16,863 mothers of children aged less than one year were interviewed, corresponding to a 3.5% to 4.4% increase in the sampling error while maintaining all other sample-size calculation criteria fixed.

A pretested form with closed questions on demographic and socioeconomic characteristics was used for collecting the following data; prenatal, delivery, and puerperium care; and child's growth and feeding; among others.

The present study included only the mothers of children aged less than six months (n=9,090); 30 were excluded because of missing data about the child's diet so only 9,060 forms were valid. The study sample size conforms to the sampling error criteria of 2.2, confidence coefficient of 95.0%, deff=2.0, considering the real prevalence of the study outcome (exclusive breastfeeding prevalence of 39.9%), allowing stratification by region.

The study dependent variable was exclusive breastfeeding, classified as yes or no. The child was considered to be exclusively breastfed when the child was fed only breast milk and no other fluid or solid, as recommended by the WHO¹. The 24-hour dietary recall was used to identify the breastfeeding and/or feeding practices. The questionnaire included questions about the intake of breast milk, other types of milk, and other foods, including water, tea, and other fluids in the previous 24 hours. Thus, the instrument indicates whether the child was exclusively breastfed in the 24 hours before the interview^{15 , 16}.

The independent variables were child's characteristics (age, gender, and birth weight); mother's sociodemographic characteristics (age, education level, skin color, and location of residence); prenatal care characteristics (attending prenatal care, prenatal care location, and advice on breastfeeding during prenatal care); delivery (type of delivery and delivery location); puerperium care (breastfeeding in the first hour of life, rooming-in, and recent visit from a health agent or Family Health Strategy agent).

The study sample distribution was assessed according to sociodemographic, and prenatal, delivery, and puerperium care variables (total and by region) with a Confidence Interval of 95% (95%CI). The exclusive breastfeeding prevalence was calculated for each month of the child's age with a 95%CI for all sociodemographic, prenatal, delivery, and puerperium variables, separately and by region, according to the Chi-square test (χ²). The exclusive breastfeeding Prevalence Ratio (PR) and respective 95%CI were calculated and stratified by region (Legal Amazon and Northeast region) by Poisson Regression adjusted for all variables. The significance level was set at 5% (p<0.05).

All estimates take into consideration the study design, correcting for the clustering effect and giving individual weights proportional to sampling probability. The weighting factor according to the number of children aged less than one year of each municipality, according to the 2010 Census, was also considered (http://www.ibge.gov.br). The data were treated by the software Stata version 11.0, using the svy command, with a significance level of 5% (p<0.05) and 95%CI.

RESULTS

Of the 9,060 study children, 4,116 were from the Legal Amazon and 4,944 were from the Northeast regions; both groups had similar ages and gender distribution. Roughly 94% of the children in both regions had normal birth weight.

Most mothers from both regions were 20 to 34 years old, finished high school, were Black, lived in urban areas other than the capital, attended prenatal care at a public service, received breastfeeding advice during prenatal care, delivered at a public service, breastfed in the first hour of life, and were roomed-in with their child. In the Legal Amazon, most of the study mothers had not been visited by the health agent or Family Health Strategy agent, unlike the Northeast mothers who had (Attached 1).

The prevalence of exclusive breastfeeding on the first month of life was 72.0%, decreasing to 57.5%, 49.9%, 39.5%, 24.4%, and 11.6% from the second to the sixth months in the study Legal Amazon municipalities. In the Northeast region, exclusive breastfeeding also decreased, reaching 13.3% at six months of age but starting at 66.3%. The prevalence of exclusive breastfeeding in the Northeast sample was statistically lower than that in the Legal Amazon sample, but equal when compared on a monthly basis (Figure 1).

Figure 1 Distribution of the prevalence of exclusive breastfeeding and confidence interval of 95% according to the child's age in months. Children younger than six months of age living in the study municipalities of the Legal Amazon and Northeast regions. Brazil, 2010. 

In the Legal Amazon municipalities, the prevalence of exclusive breastfeeding was greater among older women, reaching 48.1% in mothers aged more than 35 years; in those who breastfed the child in the first hour of life; and in those who had not been visited recently by a health agent or Family Health Strategy agent (p<0.05). In the Northeast municipalities, the prevalence of exclusive breastfeeding increased with mother's age, reaching 45.6% in the mothers aged >35 years; in those who attended prenatal care; and in those who had been visited recently by a health agent or Family Health Strategy agent (p<0.05) (Attached 2).

Multivariate analysis identified the child's age as the main variable associated with early weaning in both regions (Attached 3). The mother's age comes next: children whose mothers are older than 34 years are 28% and 42% more likely to be exclusively breastfed in the Legal Amazon and Northeast regions, respectively, than those of mothers aged less than 20 years. Living in the State Capital (17%) and breastfeeding in the first hour of life (16%) were protective factors only in the Legal Amazon.

DISCUSSION

The objective of this study was to investigate the factors associated with exclusive breastfeeding in children aged less than six months from primary municipalities in the Legal Amazon and Northeast regions of Brazil, since the literature has indicated that early weaning is strongly influenced by geographic region¹⁷. This situation was confirmed by the present study: exclusive breastfeeding was associated with different factors in the two study regions. The variables associated with exclusive breastfeeding were: child's age, mother's age, location of residence, and breastfeeding in the first hour of life, but the associations varied by region.

The prevalence of exclusive breastfeeding in children aged less than six months found by the present study (39.9%) was similar to that found by the II Breastfeeding Prevalence Survey (41.0%), including variation by region¹⁶. In the present study, the prevalence of exclusive breastfeeding in the selected Northeast municipalities was lower in children aged less than six months, but equal to that of the Legal Amazon when compared by month, declining sharply with the child's age. The decreasing prevalence of exclusive breastfeeding with child's age was also observed by Pereira et al. ¹⁸ in a study of 1,029 mothers of children aged less than six months frequenting primary health care units in the city of Rio de Janeiro: the prevalence of breastfeeding decreased by 17.0% per month of the child's life.

One can see that, in Latin American countries, exclusive breastfeeding duration varies greatly. Data from demographic and health surveys show that the median breastfeeding duration of four months in Bolivia was much greater than the medians in Brazil (1.4 months), El Salvador (1.4 months), Dominican Republic (0.5 months), and Haiti (0.4 months), countries with the lowest medians^{11 , 19}.

The mother's age was strongly associated with exclusive breastfeeding in both regions. The prevalence of exclusive breastfeeding was greater among older mothers even after adjustment: higher mother's age was a protective factor for exclusive breastfeeding. These results are similar to those found by Venâncio & Monteiro²⁰ in a study of 34,435 children younger than six months of age from 111 municipalities in the state of São Paulo, which found that the duration of exclusive breastfeeding increased with mother's age until the age group of 25 to 29 years. Saldiva et al. ¹⁷ studied 18,929 children aged less than six months from the II National Breastfeeding Prevalence Survey and found that adolescent mothers (<age 20 years) were more likely to introduce porridge early.

Breastfeeding within the first hours after birth is important for breastfeeding maintenance, and is a recommendation of the Pan-American Health Organization²¹. After adjustment for all variables included in multiple analysis, the variable breastfed within the first hour after birth' remained in the model as a protective factor for children aged less than six months from the Legal Amazon region and the study state capitals.

Santos et al. ²² studied the factors associated with preventable deaths in children aged less than four years in a cohort born in Rio Grande do Sul and found that not being breastfed within the first 24 hours of life was associated with a higher risk of dying from preventable causes, even after adjustment for confounders. Not being breastfed in the first 24 hours of life reduces the prevalence of breastfeeding^{23 , 24}. Although marginally, breastfeeding within the first hour of life is conceptually important and according to many studies, promotes breastfeeding. Therefore, promoting breastfeeding in the first 24 hours after delivery it is an essential prenatal care activity.

In adjusted data from household surveys on mother mother/child nutrition in the State of Pernambuco, Carminha et al. ²⁵ found that living in the Metropolitan Region of Recife promoted exclusive breastfeeding. Living in the state capital strongly indicates access to health services and strategies that promote and encourage breastfeeding. On the other hand, Ramos et al. ²³ and Demétrio et al. ²⁶ pointed out that living in an urban area hinders breastfeeding.

Integrated actions covering prenatal care, delivery care, and post-delivery support work together to improve the quality of care provided to women who breastfeed²⁷.

Demétrio et al. ²⁶ in a cohort from municipalities in the Recôncavo Region of Bahia, (BA) found that the median duration of exclusive breastfeeding is proportionally smaller in women who do not attend prenatal care. During prenatal care health professionals have a positive influence on breastfeeding duration, both as breastfeeding educators and promoters. In this study, the Northeast region presented the highest prevalence of exclusive breastfeeding among mothers who attended prenatal care. However, after adjustment for the confounders, this association lost significance.

According to the WHO, health agents increase exclusive breastfeeding duration more effectively than any other professional²⁸. However, the present study found that exclusive breastfeeding prevalence was higher in both regions among mothers who had not been recently visited by Family Health Strategy professionals. After adjustment in multiple analysis, this factor was not associated with breastfeeding prevalence. Studies that assessed Family Health Strategy teams' knowledge about breastfeeding found that although the professionals were well informed about the advantages of breastfeeding, few knew how to manage breastfeeding appropriately in the clinical setting^{29 - 31}. Hence, one has to assess the efficacy of these actions, since so far they have not been enough to promote exclusive breastfeeding in the two study regions.

The main variable associated with exclusive breastfeeding was child's age: in the second month, there was a 20% risk of introducing other foods; in the sixth month, the risk increased to 80%. Considering that the country has been advancing breastfeeding since 1981 with different breastfeeding-promoting actions, the data shown herein are concerning despite the increase in exclusive breastfeeding duration from 23.4 days to 54.1 days between 1999 and 2008¹⁶. The Neonatal Call³² identified a median exclusive breastfeeding duration of 64 days (76 in the Legal Amazon and 58 in the Northeast) in children aged less than six months, confirming the exclusive breastfeeding prevalence curves shown in Attached 1.

However, this type of study is not enough to identify the causes of introducing foods early. The question is: why do mothers introduce foods early? Is it a cultural and/or social issue or is it encouraged by health professionals and/or services? Many variables associated with early weaning are cited in the literature and should be assessed by other types of studies, such as: lack of information about the importance of breastfeeding, overestimating the benefits of weaning foods, maternal work, cultural habits, and social representations, among others.

The main limitations of this study are: its cross-sectional nature does not allow differentiating between cause and effect; and the fact that data collection used a method subject to memory bias, despite its recognized advantages (the WHO warns that the 24 hours can overestimate the proportion of exclusively breastfed children since some children are given other fluids occasionally)^{15 , 33}. The fact that the survey was conducted on the National Immunization Day (D-day) may also have been a limiting factor because it was not possible to survey a high percentage of the population on a single day: the vaccination period was extended, especially in hardly accessible regions, but this situation was somewhat remediated because surveys on vaccination day are still a widely used and recommended strategy in Brazil³⁴.

Although the study findings may represent the children from the study municipalities and other similar children, the 2.2% sampling error and 95.0%CI does not allow extrapolation to all children in the Brazilian Legal Amazon and Northeast regions.

CONCLUSION

In a heterogeneous country like Brazil, one should expect different regional breastfeeding practices, and this study confirmed that the factors associated with exclusive breastfeeding are influenced by the geographic region of the country. The factors associated with exclusive breastfeeding in children aged less than six months were child's age, mother's age, residence location, and being breastfed in the first hour of life, but distinctively by region.

Higher child's age was a risk factor for non-exclusive breastfeeding in both regions. In the study Legal Amazon municipalities, the probability of exclusive breastfeeding was greater among mothers aged 35 years or more. A similar situation was observed in the Brazilian Northeast. Breastfeeding in the first hour after birth and living in the state capital promote exclusive breastfeeding only in the Legal Amazon region.

These findings may be useful for defining differentiated breastfeeding promotion strategies according to the most common characteristics of each region. Strategies that should be assessed for promoting exclusive breastfeeding are the prenatal, delivery, and puerperal care that aim to reduce the early feeding of fluids or foods to children aged less than six months and to improve the processes that change hospital routines in the main maternity hospitals of the Legal Amazon and Northeast regions. Moreover, health professionals need to be better qualified with respect to managing exclusive breastfeeding until age six months as recommended by the WHO and MH, because of the strong reduction in the prevalence of exclusive breastfeeding.