Childbearing in adolescence: intergenerational dejà-vu? Evidence from a Brazilian birth cohort

This was a prospective observational study of the cohort type involving liveborns in the city of Ribeirão Preto, evaluated from June 1, 1978 to May 31, 1979, with the objective of studying perinatal factors related to baby health at birth and to infant mortality [22]. The reference population consisted of women residing in the same municipality who had delivered a live newborn at all 8 maternity hospitals (1st generation) and their daughters evaluated in 2002/04 at 23–25 years of age (2nd generation). In this second phase, we evaluated factors that, operating at the beginning of life and during the subsequent years, are related to adult health [23]. Newborn whose mothers did not reside in the city were excluded, with a total of 6973 liveborns being left, 6827 singletons and 146 twins. Twins and 343 subjects who died before 20 years of age were excluded, with 6484 subjects (males and females) being left. One of each 3 subjects belonging to the same geographic area was invited to participate in the follow-up study. In the traced group, losses to follow-up (N=705) occurred because of refusal to participate, imprisonment, death after 20 years of age, or failure to attend the interview. Losses were replaced using the same sampling frame, resulting in 2063 young adults [23], corresponding to 31.8% of the eligible population, 1068 of them being women. A total of 1059 mother-daughter pairs were available for the study, with information being available about age at first childbirth of the daughter, if this was the case, and age at first childbirth of the mother (Figure 1). Women who delivered their first child before 20 years of age were considered to be adolescent mothers. The mothers of the original study of the birth cohort (1978/79) are referred to as “1st generation” and their daughters evaluated during adult age (2002/04) as “2nd generation”.

A sample size of 1,059 has a 90% power to detect an RR of 1.7, assuming the event has a 10% prevalence in the control group, with a 5% probability of type I error.

The study was approved by the Research Ethics Committee of the University Hospital, Faculty of Medicine of Ribeirão Preto, University of São Paulo (protocol 7606/99) and all subjects gave written informed consent to participate. Details of the study methodology are available in other publications [22‐24].

We used a questionnaire with sociodemographic and biological information applied to 1059 2nd generation women, in addition to the questionnaire applied to their mothers in 1978/79 (1st generation). The following variables concerning the 1st generation were recorded: age at first childbirth in complete years, schooling in years of study (up to 4, 5 to 8, 9 or more); occupation of family head based on the International Classification of Occupation [25] (non-manual, skilled and semi-skilled manual, unskilled manual and outside the economically active population - EAP). For the 2nd generation the following information was recorded: age at first childbirth in complete years, when it occurred, schooling in years of study (up to 8, 9 or more) and occupation of family head.

Information on ethnicity was collected as the National Census does, i.e., self-reported skin colour and was categorized as white and non-white. Brazil has the largest African origin population outside Africa and presents a substantial race inequality. Official data suggest that in the period this study was done non-white male workers earned 63% less income than white ones [26]. This gap is mainly due to the advantages of whites in human capital, both own and parental education [27].

In 2002–04 retrospective information was collected regarding the adolescence period of the second generation: age at menarche (<12, 12, >12 years) and age at first job in complete years (<14, 14 to 17, > 17 or never worked).

According to the Brazilian Institute of Geography and Statistics in the 90’s most adolescent workers' (65%) came from families with an income of up to 1 minimum wage per capita [28]. The International Labour Organization reports as main causes for adolescent work the lack of access to education, poverty, specific vulnerability (low income or monoparental families and families with disabled people). Girls have a greater likelihood of receiving smaller wages, of starting to work earlier and of having the triple burden of a paid job, the domestic works and continuing to study [29]. For these reasons we chose to consider the age at first job as a proxy of the socioeconomic status during the adolescent years of the second generation, those of lower status tending to start earlier.

In 2002/04 schooling was categorized differently because in 1978–79 national compulsory schooling was 4 years and in 2002–04 it was already 8 years. In this sense lower categories in both generations represent the compulsory schooling. The highest category (9 or more years) was identical in both periods and it was chosen as the reference one.

Childbearing rates during adolescence were calculated for the two generations. Poisson regression models with robust estimate of variance were constructed to determine the independent effect of adolescent childbearing in the 1st generation on the outcome (adolescent childbearing in the 2nd generation). The incidence rate ratios (IRR) and their respective 95% confidence intervals (95% CI) were calculated.

The collection of two socioeconomic variables, occupation and schooling, was repeated at an interval of 23/24 years.

The multivariable model was adjusted for the socioeconomic variables of 1978/79 and 2002/04 as well as for age at menarche and age at first job. The linearity test was applied to investigate the presence of a gradient in the effect of exposure.

The STATA 10.1® statistical package was used for all analyses and the level of significance was set at 5%.