Adolescent pregnancy and completion of basic education : a study of young people in three state capital cities in Brazil

This study evaluated the association between adolescent pregnancy and the completion of basic education, mediated by macrosocial indicators. A cross-sectional household survey was conducted with individuals between the ages of 18 and 24 in three Brazilian cities. For the purposes of this study, individuals between the ages of 20 and 24 were selected from this sample survey that included 4,634 people. A total of 29.6% of the girls declared that they had become pregnant prior to reaching the age of 20, while 21.4% of the boys stated that they had made a girl pregnant in adolescence. Girls from households with a per capita family income of US$70 or less and who became pregnant at least once during adolescence were more likely to have not completed basic education; whereas from households with a per capita family income of US$70 or less, with parents who separated before the adolescent reached the age of 20 and that had made a partner pregnant prior reaching the age of 20 were more likely to have not completed basic education. It is vital that the school system provides girls and boys with guidance on sexuality and contraception and encourages them to remain in education. Adolescent Pregnancy; Gender Identity; Education Introduction The past few decades have seen changes in the transition from childhood to adulthood, with a prolonged youth phase, postponed parenthood and young people tending to remain in education for longer periods of time. Nevertheless, adolescent pregnancy remains a possibility in the trajectory of young people 1 and has been associated with poor academic performance and higher rates of school dropout. Adolescent pregnancy is also attributable to low social economic status, which may be aggravated when the pregnancy occurs in the absence of a formal union 2. Many studies of the relationship between adolescent pregnancy and education levels have been published 2,3,4,5,6,7,8,9,10. In Brazil, studies have shown that girl’s education level is the strongest influencing factor and consistently associated with adolescent pregnancy 9. In addition, most pregnant teenagers from the less privileged classes of society and poorer regions of the country drop out of school before completing their basic education 4. The term basic education here refers to elementary and high school. These studies have also shown higher rates of illiteracy among 17 to 19 year old adolescent mothers and indicate that adolescent pregnancy is more likely among girls with lower levels of education 10. The economic and social consequences of becoming pregnant before reaching the age of 2386 ARTIGO ARTICLE ADOLESCENT PREGNANCY AND COMPLETION OF BASIC EDUCATION 2387 Cad. Saúde Pública, Rio de Janeiro, 27(12):2386-2400, dez, 2011 20 vary according to culture and family structure 2,11. Middle and upper class teenagers generally receive financial and material support, thereby reducing the impact of teenage parenthood 12 on the educational and professional future of the individual. Furthermore, while abortion is more common among middle and upper class teenagers, less privileged teenagers generally carry the pregnancy to term, often negatively affecting their education 13. Education in Brazil is marked by student age/ grade gaps; i.e., in many cases older children are mixed with younger children in the same class, especially in the lower grades. This is due to a number of factors such as school drop out, grade repetition, and children starting to attend school after the required age, which invariably result in further school dropouts. In addition to macrosocial determinants, studies indicate a relationship between family context and academic performance 14,15,16,17,18. Despite the fact that access to education and the opportunity to remain in school are initially similar for both sexes, girls remain in education for longer than boys 19,20. Gender inequalities become evident when factors such as the number of young girls who interrupt their education, repeat grades and fail to complete basic education, principally when a reproductive event occurs, are taken into consideration 8. In measuring the relationship between school dropout rates and pregnancy before reaching the age of 20, Almeida et al. 8 reported that 36.4% of girls and 40.4% of boys abandoned their studies either during or following a pregnancy. Among adolescents from households with a per capita family income of one minimum salary or less, around 20% of girls and boys had already dropped out of school before the pregnancy occurred. In the case of the boys, the principal reason given for having left school was work-related, while the girls mentioned that the main motive was pregnancy and having to take care of the child. Given the complexity of this phenomenon, investigating the effect of adolescent pregnancy on the completion of basic education is not an easy task; however, it is an important one. This paper uses data obtained by the GRAVAD study – a multicenter study of young people, sexuality and reproduction in Brazil 21. The objective of this paper is to report findings on the association between adolescent pregnancy and completion of basic education among the individuals interviewed, mediated by macro-social indicators. Methodology This cross-sectional study was performed in 2001 with young people from Porto Alegre (capital of the Rio Grande do Sul State), Rio de Janeiro (capital of the Rio de Janeiro State) and Salvador (capital of the Bahia State). The participants were between 18 and 24 years of age at the time of the study. It was decided to use a probabilistic sample stratified in three stages in order to guarantee representativeness of the general population in the targeted age-group. Calculations indicated a required sample size of 1,500 individuals for each city, based on estimated pregnancy rates among 18 and 19 year-old girls of 25.3% and 34.8% respectively, based on statistics from 1996 22. Geographical dispersion was guaranteed by the sampling design. The first step involved a randomized sample stratified by census area and grouped into five strata based on the mean income and education level of the head of the family. In the second step, 33 of the households in each census sector, previously identified as being eligible because they contained an individual aged between 18 and 24 years of age, were independently randomized, allowing for a possible loss rate of around 20%. In the third step, the person whose birthday occurred earliest in the calendar year was chosen from the eligible individuals in each selected household to participate in the study. In the case of identical birthdates, the individual whose name came first in the alphabet was selected for inclusion in the study. Data was collected through face-to-face interviews using a previously tested questionnaire to obtain information on academic trajectories, employment and affective and sexual relationships. Young people studying or working in areas related to human sciences, who were approximately the same age as the participants in the study, were trained to conduct the interviews. To guarantee the quality of the data collected, the questionnaires were reviewed by a team of supervisors. In addition, 20% of the questionnaires were randomly selected and tested by reasking specific questions using a telephone interview or by returning to the individual’s home. The database was constructed using the Epi Info software package, version 6.04b (Centers for Disease Control and Prevention, Atlanta, USA), using double data entry procedures. In view of the complexity of the sample design, the data was analyzed using the Stata software program, version 8 (Stata Corp., College Station, USA), which permitted the incorporation of sample design effects and the relative weight of each unit, thus obtaining reliable parameter estimates. Weight Almeida MCC, Aquino EML 2388 Cad. Saúde Pública, Rio de Janeiro, 27(12):2386-2400, dez, 2011 was calculated based on the probability of the selection of the units in each step. In constituting the study population, it was decided to redefine the age-group usually defined as adolescent 23,24 so as to permit the inclusion of any pregnancy that had occurred previously but that was sufficiently recent for the individual to remember the details relevant to the study. This study analyzed data from individuals between 20 and 24 years of age who were interviewed, irrespective of whether they had been pregnant or not. The fact that all these individuals had already passed from adolescence to adulthood and therefore were not able to experience any further reproductive episodes in adolescence, together with the fact that the participants should technically have already completed their basic education, were factors that were taken into account when defining the study population. An individual was considered to be outside the appropriate age-group for a particular school grade (age-grade gap) if he/she was over seven years of age when he/she started to attend elementary school. The principal variables selected for the study were those relating to schooling: late starting age in attending elementary education (yes/ no); completed basic education (yes/no); and at least one reported pregnancy in adolescence (defined as having occurred before reaching the age of 20). The covariables of interest were related to macrosocial and family characteristics: city of residence, per capita monthly family income, skin color/ethnic group (self-reported), mother’s education level, number of siblings, whether the individual’s parents were separated, participation in domestic chores in adolescence, and whether the individual worked in paid employment for more than three months prior to turning 20. All the covariables were dichotomized to perform multivariate analysis with the exception of city of residence. The first step of the analysis procedures consisted of describing the selected variables using the distributi


Introduction
The past few decades have seen changes in the transition from childhood to adulthood, with a prolonged youth phase, postponed parenthood and young people tending to remain in education for longer periods of time.Nevertheless, adolescent pregnancy remains a possibility in the trajectory of young people 1 and has been associated with poor academic performance and higher rates of school dropout.Adolescent pregnancy is also attributable to low social economic status, which may be aggravated when the pregnancy occurs in the absence of a formal union 2 .
Many studies of the relationship between adolescent pregnancy and education levels have been published 2,3,4,5,6,7,8,9,10 .In Brazil, studies have shown that girl's education level is the strongest influencing factor and consistently associated with adolescent pregnancy 9 .In addition, most pregnant teenagers from the less privileged classes of society and poorer regions of the country drop out of school before completing their basic education 4 .The term basic education here refers to elementary and high school.These studies have also shown higher rates of illiteracy among 17 to 19 year old adolescent mothers and indicate that adolescent pregnancy is more likely among girls with lower levels of education 10 .
The economic and social consequences of becoming pregnant before reaching the age of Cad.Saúde Pública, Rio de Janeiro, 27(12):2386-2400, dez, 2011 20 vary according to culture and family structure 2,11 .Middle and upper class teenagers generally receive financial and material support, thereby reducing the impact of teenage parenthood 12 on the educational and professional future of the individual.Furthermore, while abortion is more common among middle and upper class teenagers, less privileged teenagers generally carry the pregnancy to term, often negatively affecting their education 13 .
Education in Brazil is marked by student age/ grade gaps; i.e., in many cases older children are mixed with younger children in the same class, especially in the lower grades.This is due to a number of factors such as school drop out, grade repetition, and children starting to attend school after the required age, which invariably result in further school dropouts.In addition to macrosocial determinants, studies indicate a relationship between family context and academic performance 14,15,16,17,18 .
Despite the fact that access to education and the opportunity to remain in school are initially similar for both sexes, girls remain in education for longer than boys 19,20 .Gender inequalities become evident when factors such as the number of young girls who interrupt their education, repeat grades and fail to complete basic education, principally when a reproductive event occurs, are taken into consideration 8 .
In measuring the relationship between school dropout rates and pregnancy before reaching the age of 20, Almeida et al. 8 reported that 36.4% of girls and 40.4% of boys abandoned their studies either during or following a pregnancy.Among adolescents from households with a per capita family income of one minimum salary or less, around 20% of girls and boys had already dropped out of school before the pregnancy occurred.In the case of the boys, the principal reason given for having left school was work-related, while the girls mentioned that the main motive was pregnancy and having to take care of the child.
Given the complexity of this phenomenon, investigating the effect of adolescent pregnancy on the completion of basic education is not an easy task; however, it is an important one.
This paper uses data obtained by the GRAVAD study -a multicenter study of young people, sexuality and reproduction in Brazil 21 .The objective of this paper is to report findings on the association between adolescent pregnancy and completion of basic education among the individuals interviewed, mediated by macro-social indicators.

Methodology
This cross-sectional study was performed in 2001 with young people from Porto Alegre (capital of the Rio Grande do Sul State), Rio de Janeiro (capital of the Rio de Janeiro State) and Salvador (capital of the Bahia State).The participants were between 18 and 24 years of age at the time of the study.It was decided to use a probabilistic sample stratified in three stages in order to guarantee representativeness of the general population in the targeted age-group.Calculations indicated a required sample size of 1,500 individuals for each city, based on estimated pregnancy rates among 18 and 19 year-old girls of 25.3% and 34.8% respectively, based on statistics from 1996 22 .
Geographical dispersion was guaranteed by the sampling design.The first step involved a randomized sample stratified by census area and grouped into five strata based on the mean income and education level of the head of the family.In the second step, 33 of the households in each census sector, previously identified as being eligible because they contained an individual aged between 18 and 24 years of age, were independently randomized, allowing for a possible loss rate of around 20%.In the third step, the person whose birthday occurred earliest in the calendar year was chosen from the eligible individuals in each selected household to participate in the study.In the case of identical birthdates, the individual whose name came first in the alphabet was selected for inclusion in the study.
Data was collected through face-to-face interviews using a previously tested questionnaire to obtain information on academic trajectories, employment and affective and sexual relationships.Young people studying or working in areas related to human sciences, who were approximately the same age as the participants in the study, were trained to conduct the interviews.
To guarantee the quality of the data collected, the questionnaires were reviewed by a team of supervisors.In addition, 20% of the questionnaires were randomly selected and tested by reasking specific questions using a telephone interview or by returning to the individual's home.
The database was constructed using the Epi Info software package, version 6.04b (Centers for Disease Control and Prevention, Atlanta, USA), using double data entry procedures.In view of the complexity of the sample design, the data was analyzed using the Stata software program, version 8 (Stata Corp., College Station, USA), which permitted the incorporation of sample design effects and the relative weight of each unit, thus obtaining reliable parameter estimates.Weight was calculated based on the probability of the selection of the units in each step.
In constituting the study population, it was decided to redefine the age-group usually defined as adolescent 23,24 so as to permit the inclusion of any pregnancy that had occurred previously but that was sufficiently recent for the individual to remember the details relevant to the study.This study analyzed data from individuals between 20 and 24 years of age who were interviewed, irrespective of whether they had been pregnant or not.The fact that all these individuals had already passed from adolescence to adulthood and therefore were not able to experience any further reproductive episodes in adolescence, together with the fact that the participants should technically have already completed their basic education, were factors that were taken into account when defining the study population.An individual was considered to be outside the appropriate age-group for a particular school grade (age-grade gap) if he/she was over seven years of age when he/she started to attend elementary school.
The principal variables selected for the study were those relating to schooling: late starting age in attending elementary education (yes/ no); completed basic education (yes/no); and at least one reported pregnancy in adolescence (defined as having occurred before reaching the age of 20).
The covariables of interest were related to macrosocial and family characteristics: city of residence, per capita monthly family income, skin color/ethnic group (self-reported), mother's education level, number of siblings, whether the individual's parents were separated, participation in domestic chores in adolescence, and whether the individual worked in paid employment for more than three months prior to turning 20.All the covariables were dichotomized to perform multivariate analysis with the exception of city of residence.
The first step of the analysis procedures consisted of describing the selected variables using the distribution of simple frequencies and cross-analysis in order to characterize the study population.The differences in proportions were tested for statistical significance using Pearson's chi-square test with Rao and Scott's second order correction at the 5% significance level.However, it was decided to present the 95% confidence intervals (95%CI).
Non-conditional logistic regression models were constructed separately for boys and girls for the variable completed basic education (yes/no) which, as mentioned above, included all levels of education except university.
For the purpose of this analysis, non-completion of basic education was defined as a dependent variable.The covariables shown by the bivariate analysis as having an association with the completion of basic education (with p-values ≤ 0.20) were selected for inclusion in the regression models.
The strategy used for including the variables in the models was hierarchical and considered four blocks of covariables: in the first hierarchical level, the covariables representing the macrosocial factors were introduced; in the following step, the remaining variables associated with failure to complete basic education, after simultaneous adjustment (with p-levels < 0.05), were maintained; the same procedure was applied to the microsocial variables relating to family context, paid employment and finally the occurrence of pregnancy before reaching the age of 20.These variables were selected because they were considered most relevant by the young people themselves.
The study was conducted in compliance with all applicable ethical procedures and the protocol was approved by the Internal Review Boards of the respective universities involved in this work [the Federal University of Bahia (Universidade Federal da Bahia -UFBA), the State University of Rio de Janeiro (Universidade do Estado do Rio de Janeiro -UERJ), and the Federal University of Rio Grande do Sul (Universidade Federal do Rio Grande do Sul -UFRGS)].An informed consent form was read by each participant prior to the questionnaire and signed following its completion.In addition, the privacy of the individual was guaranteed by ensuring that the interview took place without the presence of any third parties over four years of age.Furthermore, the individual participated in the process on a voluntary basis and the participant's right to refuse to answer any question or to stop the interview at any stage was respected.

Results
Of 4,634 individuals interviewed under the GRAVAD study, 3,042 (65.6%) young people who were between the ages of 20 and 24 at the time of the interview were included in the present study.Of this total, 53.2% were male and 46.8% female.Overall, 29.6% of the girls reported that they had been pregnant and 21.4% of the boys mentioned that they had made a girl pregnant before reaching the age of 20.A total of 8.8% of the girls and 15.4% of the boys reported that they were over seven years of age when they began to attend school (i.e., there was already an age/grade gap when they started school).Approximately 81% of the girls and 75% of the boys completed elementary school.In contrast, only 61.2% of the girls and 50.6% of the boys reported having completed high school (data not shown).
A total of 14.3% of the girls who became pregnant during adolescence were over seven years of age when they started school.This percentage was almost twice that found in the girls who had not become pregnant (Table 1).With respect to the girls who had not become pregnant during adolescence, different results were found by city.The age/grade gap for girls was found to be greatest in Salvador and in this city the percentage of girls with an age/grade gap was twice as high in girls who had become pregnant in adolescence (Table 1).This age/grade gap was more common in girls from households with lower per capita family income, irrespective of whether the girl had become pregnant in adolescence or not.Likewise, girls whose mothers had failed to complete elementary school were more likely to be outside the age-group for their school grade; However, in the group of girls who had become pregnant in adolescence, the likelihood of their mothers having failed to comlete elementary school was even greater (Table 1).Among the group of girls with an age/grade gap, no statistically significant differences in color/ethnicity were found between girls who had become pregnant in adolescence and girls that had not become pregnant.However, the proportion of an age/grade gap was greater among girls who had reported at least one pregnancy in adolescence, particularly in black girls.
The same relationship was found with respect to the covariables whether their parents were separated, participation in household chores in adolescence and whether they worked in paid employment for more than three months prior to reaching the age of 20.However, the proportion of an age/grade gap in girls with two or more siblings was approximately three times greater than in girls with only one sibling and twice as great compared to girls who did not become pregnant in adolescence (Table 1).A total of 70.5% of the girls who reported at least one pregnancy in adolescence failed to complete basic education compared to 25.6% of the girls who had not become pregnant (Table 1).The proportion of girls that did not become pregnant in adolescence who failed to complete basic education was highest in Salvador; however, these differences between cities were cancelled out by the results related to the occurrence of pregnancy.
The effect of family income and mother's education level on completion of basic education was aggravated by pregnancy in adolescence (Table 1).The percentage of girls that did not complete basic education was higher among girls with two or more siblings and this disadvantage increased when the pregnancy occurred before the girl reached the age of 20.The fact that the adolescent's parents separated before the girl reached the age of 20 and the fact of having worked during adolescence did not affect school dropout rates within the groups.However, the proportion of non-completion of school in girls who became pregnant and whose parents were separated was greater than in girls who had not become pregnant in adolescence.
With respect to participation in household chores, in both groups (girls who became pregnant during adolescence and girls that did not), no significant differences in non-completion of school rates were found between the girls who only helped out around the house or had no obligations at all and those who were completely responsible for household chores (Table 1).
A total of 25.9% of the boys who made a girl pregnant during adolescence were over seven years of age when they began school, compared to 12.5% of boys who had no involvement in pregnancy during adolescence (Table 2).In the group of boys who had no involvement in pregnancy during adolescence, the proportion of an age-grade gap was highest in Salvador.However, in the group of boys involved in pregnancy, the proportion of an age-grade gap in boys in Salvador was similar to that found among boys in Rio de Janeiro.It should also be noted that the proportion of an age-grade gap is 20 times greater in boys from households with lower per capita family income compared to boys from households with with a higher per capita family income.This difference was considerably attenuated in the group of boys involved in pregnancy in adolescence.The mother's education level does not appear to have any effect on age-grade gap in boys; however, the proportion of an age-grade gap increases in boys involved in pregnancy in adolescence (Table 2).Having more siblings was found to be a disadvantage for boys with an age-grade gap irrespective of whether they made a girl pregnant in adolescence or not (Table 2).
With respect to the non-completion of basic education, the boys repeated the same pattern as the girls (Table 2); however, among the boys not involved in pregnancy, non-completion rates are generally twice those found among the girls across all levels of the analysis.
In girls, a strong crude association was found between non-completion of basic education and the mother having a low education level (OR = 8.30; 95%CI: 5.41-12.74),coming from a household with lower per capita family income (OR = 6.59; 95%CI: 4.60-9.45)and pregnancy in adolescence (OR = 7.16; 95%CI: 5.07-10.12).A weaker association was also found between non-completion of basic education and the following factors: self-reported skin color/ethnicity, number of siblings, the fact that parents had separated and participation in household chores (Table 3).
At the first level, coming from a household with low per capita family income and the girl's mother having a low level of education, were factors associated with non-completion of high school.At the second level, the covariables related to the family context -number of siblings, separated parents and participation in household chores -were found to have a statistically significant association and were therefore maintained in the model.No significant association was found between the covariable paid employment and non-completion of high school.At the fourth level, after adjusting the final model by including adolescent pregnancy, the covariables separated parents and participation in household chores were found to lose statistical significance.Once again, even after adjustment for the other variables, a strong association was found between adolescent pregnancy and the girl's mother having a low level of education and non-completion of basic education (Table 3).
A strong association was found between the failure of boys to complete basic education with the fact that boys came from a household with lower per capita family income (OR = 9.18; 95%CI: 5.98-14.12)and with the boy's mother having a low level of education (OR = 6.93; 95%CI: 5.00-9.63).Associations of lesser magnitude were also found with skin color/ethnicity, having more siblings, having separated parents, participation in household chores and being in paid employment for more than three months prior to reaching the age of 20 (Table 4).With respect to macrosocial factors, at the first level the variable city lost statistical significance in all categories and for this reason was removed from the model.Having two or more siblings and the fact that parents had separated prior to the individual reaching the age of 20 were factors associated with non-completion of basic education, even following adjustment for the other factors.No association was detected with being in paid employment for more than three months prior to reaching the age of 20.However, the non-completion of school rate was higher in boys who made a girl pregnant during  adolescence compared to boys with no involvement in pregnancy, even in the adjusted model.The majority of the girls in elementary or high school who became pregnant during adolescence carried their first pregnancy to term.The percentage of girls that carried their first pregnancy to term was higher (81.9%) in the case of girls that did not complete elementary school than in the case of girls who completed elementary school; however, the statistical significance of this difference was borderline.Reports of abortion were around four times higher in the girls who graduated from high school compared to girls that failed to complete high school and this difference was statistically significant (Figure 1).
Boys in elementary school and high school who had made a girl pregnant in adolescence reported higher rates of abortion; however, there were no statistically significant differences be-tween this pregnancy outcome and completion of high school.Abortion was reported twice as often as births (57.4% and 25% respectively) by boys who completed high school; however, the statistical significance of this relationship was borderline (Figure 1).

Discussion
This study was performed using data obtained from a population-based survey in three Brazilian cities carried out in a representative sample of girls and boys between the ages of 18 and 24 that have different social and cultural backgrounds.It should be emphasized that the inclusion of boys in this study provided an insight into adolescent paternity and the relationship between this phenomenon and educational achievement.Only two-thirds of the girls and half the boys interviewed completed basic education.The proportion of individuals who experienced pregnancy in adolescence that failed to complete school was very high in girls (70.5%) and boys (73.1%).For both boys and girls, coming from a household with low per capita family income and having a mother with a low level of education were factors strongly associated with non-completion of basic education.Adolescent pregnancy was also a factor associated with non-completion of school, adding to the preexisting disadvantages related to the adolescent's social and cultural background.
Intentional errors may have occurred, particularly with respect to declared pregnancy and abortion with girls underreporting pregnancy for example.However, the design of the research instrument, which included questions on the individual's reproductive life at different moments during the interview, should have ensured the identification of any inconsistencies.
Biases may have occurred in the information given by boys regarding making a girl pregnant.However, when questioning the girls who reported having become pregnant on the reaction of their partner to the pregnancy, only 2.2% stated that they did not tell him about it 25 .Although the study did not interview the girls and their respective partners, it is believed that this finding may be extrapolated to the boys who participated in the study.
The moral censorship of abortion and the fact that it is illegal in Brazil may have discouraged adolescents, particularly girls, from reporting this pregnancy outcome 26 .Although boys should have less difficulty in reporting abortion, as they were not directly involved in the procedure, it is possible that the information they provided on the girl's pregnancy is less accurate and consistent.
Considering that seven is the official school starting age for compulsory education in Brazil, the proportion of individuals who were over seven years of age when they began school (8.8% of girls and 15.4% of boys) is low in comparison to the national average of 18.9% according to the 2005 School Census on Basic Education 27 of all children enrolled in the first grade of elementary school in Brazil 28 .
It was found that the proportion of an agegrade gap at the beginning of elementary school was greater among the girls and boys who had reported pregnancy in adolescence, with the highest proportion occurring in the sample from Salvador.This finding may be a consequence of regional inequalities in education, the economy and urban development, whereby states in the north and northeast of the country have the worst education indicators (age-grade gap, repetition and dropout rates) 27 .The proportion of individuals from households with lower per capita family income and with mothers with low levels of education was also higher in this sample, indicating the selectivity of the Brazilian education system.
Although the rate of non-completion of school was greater among boys and girls who had experienced pregnancy in adolescence and who had a less favorable family background and socioeconomic status, it is important to note that a significant proportion of boys and girls with the same characteristics but that did not experience pregnancy also failed to complete their basic education.It may therefore be assumed that those in this situation were already more likely not to complete their education and that the occurrence of a reproductive event constituted an aggravating factor.For less privileged young people, pregnancy may represent a change in social status, offering better prospects for the immediate future 5 .In addition, where family conflict is involved, a partner and motherhood may seem like a solution to these problems 11 .
Among the boys, a desire for financial independence and a tendency to follow an irregular path through school can be noted, justified by the desire and need to work and loss of interest in school 29 .
Some studies attribute educational disadvantages to the type of family structure experienced by young people 7,15 .The proportion of individuals whose parents were separated at the time of pregnancy in adolescence that failed to complete basic education was greater than that of individuals whose parents remained together or who only separated after their children had gone through adolescence.However, the overlapping confidence intervals indicate that there was no statistically significant difference.This was confirmed in the groups of girls when the association between these factors lost statistical significance when other variables were introduced into the multivariate analysis.
It should be noted that this study did not evaluate the amount of time participants spent with each parent, the number of times that parents separated and the formation of other relationships by their parents.An analysis of these factors may have resulted in different interpretations of the findings, as previously described by other studies 15,16 .
The association with having more siblings was maintained at all levels of the model, suggesting that the socioeconomic status of the family has a direct effect, since when families with low income have more children it is generally necessary to share already insufficient resources among the siblings 17,30 .It is already known that there is no association between the academic performance of more privileged adolescents and the number of siblings an individual has 30 .An interesting line of investigation for future studies would be to evaluate the association between adolescent's school performance and their mother's and father's education levels, since the academic success or failure of the child appears to vary as a function of the parents' education levels 31 .
Gender inequalities were found to exist in the case of the covariable worked in paid employment for more than three months prior to turning 20.In the case of girls, work did not appear to be associated with failure to complete basic education even when pregnancy occurred in adolescence, suggesting that other motives led non-completion of school in this group.Boys who had worked for more than three months in adolescence were more likely to report having left school prior to completing basic education, irrespective of whether they had made a girl pregnant or not.However, when the relationship between work and non-completion of high school was evaluated, a crude association was found that lost statistical significance when the macrosocial and family variables were incorporated into the model.
Boys from different social groups seek work for different motives: middle class boys work to gain financial independence, whereas those from the less privileged social classes also report the need to help support the family 29,31 .Even when pregnancy in adolescence constitutes a pressing reason for an adolescent boy to seek work, it does not appear to be the principal motivating factor.On the other hand, boys from more underprivileged families report experiencing difficulties in conciliating work with study, forcing them to abandon school 31 .
The finding that more of the girls and boys who completed basic education reported abortion is in agreement with the hypothesis that these individuals prioritize an academic career over parenthood.Studies reported in the published literature show that young people from more privileged backgrounds are subjected to greater pressure from their families to complete their studies, hence postponing parenthood 5,13,29 .Similarly, young adults at university mention the use of emergency contraception so as not to hinder their future 32 .It should therefore be emphasized that the socioeconomic status of the individuals is an important factor to be considered in interpreting this finding.
The finding that more males reported abortion should be interpreted with caution.Boys find it less difficult to mention having terminated a pregnancy because girls are usually responsible for this act.The association found between pregnancy in adolescence and the non-completion of school by boys was intriguing as the principal outcome reported in these cases was abortion.This finding deserves further investigation in other studies, specifically from a longitudinal perspective that would provide a better understanding of the effect of adolescent pregnancy on the academic careers of boys.
Finally, the characteristics of the affective, sexual and reproductive trajectories experienced by boys are different from those experienced by girls.In general, boys become sexually active earlier and have more casual or simultaneous relationships, and thus they are more exposed to the risk of an unplanned pregnancy that may end up in an abortion 21 .
The objective of this study was to evaluate the association between macrosocial factors, adolescent pregnancy and the completion of basic education.Girls from households with a per capita family income of one minimum salary or less, whose mother had a low level of education, who became pregnant at least once in adolescence and who had two or more siblings were more likely to have not completed basic education; whereas boys from households with a per capita family income of US$70.00 or less, whose mother had a low level of education, with two or more siblings, having parents who separated before the adolescent reached 20 years of age and who made a girl pregnant prior to turning 20 years of age were more likely to have not completed basic education.
The results of this study indicate that the social and cultural differences of boys and girls with regard to sexuality and reproduction are a challenge to be faced by the education system.Sex education and the promotion of actions to stimulate students to remain in the education system are vital measures that need to be taken.More effective educational policies are required in order to reduce the selectivity that exists within the education system.In addition, policies should be implemented to encourage young people to return to school and these should be articulated together with policies aimed at improving the health of adolescents of both sexes, by providing guidance on the choice and use of contraceptive methods and how to gain access to them.

Figure 1 Outcome
Figure 1Outcome of the fi rst pregnancy prior to reaching the age of 20 according to whether the young person fi nished school; mean percentages completing High School according to gender in young people between the ages of 20 and 24.

Table 1
Distribution of girls between the ages of 20 and 24 that reported at least one pregnancy prior to reaching the age of 20, educational indicators and selected social and family variables.CovariablesPregnancy prior to reaching the age of 20 No pregnancy prior to reaching the age of 20 Cad. Saúde Pública, Rio de Janeiro, 27(12):2386-2400, dez, 2011 ** Categories defi ned based on the prevailing nationwide minimum wage (US$70.00).Source: GRAVAD study, 2002.Population: young people between the ages of 20 and 24 in Porto Alegre (Rio Grande do Sul State), Rio de Janeiro (Rio de Janeiro State), Salvador (Bahia State).Table 1 (continued)

Table 2
Distribution of boys between the ages of 20 and 24 that reported at least one pregnancy prior to reaching the age of 20, educational indicators and selected social and family variables.

Table 3
Hierarchical analysis using logistic regression of the factors associated with non-completion of basic education in girls between the ages of 20 and 24.
Source: GRAVAD study, 2002.Population: young people between the ages of 20 and 24 in Porto Alegre (Rio Grande do Sul State), Rio de Janeiro (Rio de Janeiro State), Salvador (Bahia State).

Table 4
Hierarchical analysis using logistic regression of the factors associated with non-completion of basic education in boys between the ages of 20 and 24.
Source: GRAVAD study, 2002.Population: young people between the ages of 20 and 24 in Porto Alegre (Rio Grande do Sul State), Rio de Janeiro (Rio de Janeiro State), Salvador (Bahia State).