Background
The risks and realities associated with teenage motherhood are well documented, with consequences starting at childbirth and following both mother and child over the life span.
Teenage births result in health consequences; children are more likely to be born pre-term, have lower birth weight, and higher neonatal mortality, while mothers experience greater rates of post-partum depression and are less likely to initiate breastfeeding [
1,
2]. Teenage mothers are less likely to complete high school, are more likely to live in poverty, and have children who frequently experience health and developmental problems [
3]. Understanding the risk factors for teenage pregnancy is a prerequisite for reducing rates of teenage motherhood. Various social and biological factors influence the odds of teenage pregnancy; these include exposure to adversity during childhood and adolescence, a family history of teenage pregnancy, conduct and attention problems, family instability, and low educational achievement [
4,
5].
Mothers and older sisters are the main sources of family influence on teenage pregnancy; this is due to both social risk and social influence. Family members both contribute to an individual’s attitudes and values around teenage pregnancy, and share social risks (such as poverty, ethnicity, and lack of opportunities) that influence the likelihood of teenage pregnancy [
6,
7]. Having an older sister who was a teen mom significantly increases the risk of teenage childbearing in the younger sister and daughters of teenage mothers were significantly more likely to become teenage mothers themselves [
8,
9]. Girls having both a mother and older sister who had teenage births experienced the highest odds of teenage pregnancy, with one study reporting an odds ratio of 5.1 (compared with those who had no history of family teenage pregnancy) [
5]. Studies consistently indicate that girls with a familial history of teenage childbearing are at much higher risk of teenage pregnancy and childbearing themselves, but methodological complexities have resulted in inconsistent findings around “parent/child sexual communication and adolescent pregnancy risk” [
10]. A review of family relationships and adolescent pregnancy risk found risk factors to include living in poor neighborhoods and families, having older siblings who were sexually active, and being a victim of sexual abuse [
10]. Research around the impact of sister’s teenage pregnancy has been limited to mostly qualitative studies using small samples of minority adolescents in the United States [
5,
11].
To our knowledge, no previous studies have examined the impact of an older sister’s teenage pregnancy on the odds of her younger sister having a teenage pregnancy, and compared this effect with the direct effect of having a mother who bore her first child before age 20. By controlling for a variety of social and biological factors (such as neighborhood socioeconomic status, marital status of mother, residential mobility, family structure changes, and mental health), and the use of a strong statistical design—propensity score matching with a large population-based dataset—this study aims to determine whether teenage pregnancy is more strongly predicted by having an older sister who had a teenage pregnancy or by having a mother who bore her first child before age 20.
Discussion
The rate differences of teenage pregnancy were similar for those whose older sister had a teenage pregnancy (40.4 per 100 - 10.3 per 100 = 30.1 per 100) and for those whose mother bore her first child before age 20 (39.4 per 100 - 13.1 per 100 = 26.3 per 100). After propensity score matching on a series of variables, the odds of becoming pregnant for a teenager were much higher if her older sister had a teenage pregnancy than if her mother had been a teenage mother. For both older sisters’ teenage pregnancy and mother’s teenage childbearing, the odds in this study are lower than those reported elsewhere; this is likely due to the larger sample size, more rigorous methods, and inclusion of important predictors.
Several examinations of family histories in the literature show older sisters to have the greatest influence on a younger sister’s odds of having a teenage pregnancy. Controlling for family socioeconomic status, maternal parenting, and sibling relationships, teens with an older sister who had a teenage birth were 4.8 times more likely to have a teenage birth themselves; these odds increased to 5.1 if both the older sister and mother had a teenage birth [
11]. Four older studies estimated the rate of teen pregnancy to be between 2 and 6 times higher for those with older sisters having a teenage pregnancy [
41]. This work focused primarily on young black women in the United States and controlled for limited confounders (aside from race and age). None of the previous studies examining the impact of an older sister’s teenage pregnancy controlled for mother’s teenage childbearing or time-varying factors before age 14 (mental health, residential mobility, family structure changes); this research probably overestimated the relationship between sisters’ teenage pregnancy status.
The mechanisms driving the relationship between an older sister’s teenage pregnancy and the pregnancy of a younger adolescent sister have been examined through approaches based on social learning theory, shared parenting influences, and shared societal risk [
41]. Bandura’s social learning theory indicates that “most human behavior is learned observationally through modeling: from observing others one forms an idea of how new behaviors are performed, and on later occasions this coded information serves as a guide for action” [
7]. When sisters live in the same environment, seeing an older sister go through a teenage pregnancy and childbirth may make this a more acceptable option for the younger sister [
11]. Not only do both sisters have the same maternal influence that may affect their odds of teenage pregnancy, having an older sister who is a teenage mother may change the parenting style of the mother. Mothers involved in parenting of their teenage daughters’ child may have “supervised their children less, communicated with their children less about sex and contraception, and perceived teenage sex as more acceptable when the older daughter’s status changed from pregnant to parenting” [
42]. Finally, both sisters share the same social risks, such as poverty, ethnicity, and lack of opportunities, that increase their chances of having a teenage pregnancy [
42].
Having a mother bearing her first child before age 20 was a significant predictor for teenage pregnancy. We found daughters of teenage mothers to be 51 % more likely to have a teenage pregnancy than those whose mothers were older than 19 when they bore their first child. This is quite close to the 66 % increase found by Meade et al (2008), who controlled for many of the same variables except having an older sister with a teenage pregnancy, and the time-varying covariates of family structure change, mental health conditions, and residential mobility. Meade et al. [
9] did adjust for school performance; in the adjusted sub-sample, the odds ratio reduced to 1.34, indicating a 34 % increase in teenage pregnancy.
Intergenerational teenage pregnancy may be influenced by such mechanisms as “biological heritability, intergenerational transmission of values regarding family, the mother’s level of fertility, the indirect impact of socioeconomic and family environment through educational deficits or low opportunity or aspirations, and directly through the mother’s role modeling” [
43]. Women bearing their first child in their adolescence are more likely to pass on “risky” characteristics, which could produce negative outcomes in their offspring [
44]. Another mechanism identified as contributing to intergenerational teenage pregnancy is that daughters of teenage mothers have an increased internalized preference for early motherhood, have low levels of maternal monitoring, and are thus more likely to become sexually active at a young age and engage in unprotected sex [
44]. The influence of a mother’s teenage pregnancy therefore works through the environment created and parenting style assumed as a result of a mother’s teenage childbearing.
The use of administrative data to conduct health services research has some significant advantages and limitations. Administrative data from a large birth cohort have higher levels of accuracy is not depending on recall (such as in retrospective surveys) and is ideal for examining risk factors over time due to the longitudinal follow-up [
45]. These data—with a large N and a number of covariates—are well-suited for propensity scoring. A significant limitation (shared with almost all observational studies) is that certain covariates and mediating effects are unobservable due to lack of information. The data can only capture recorded variables; for example, only individuals seeking mental health treatment will receive a diagnosis, which may not be include all individuals with mental health conditions [
46]. Sensitivity testing addresses this limitation, but such covariates might well have impacted study results. As mentioned above, not adjusting for involvement with child protective services (such as CFS) is a limitation. Although the number of teenage girls involved with CFS is relatively small, they may not be interacting with their mother or older sister on a regular basis and thus are less likely to model themselves after their family members. The availability of an educational predictor was an identified limitation. To account for the impact of educational achievement in our full cohort, educational outcomes would need to be available for everyone for grade 7 at the latest (as almost all teenage pregnancies occur after grade 7). Since educational achievement generally remains quite similar from year to year—grade 9 achievement is likely to be quite similar to grade 7 achievement [
30]; this reduced odds ratio may better estimate the true odds. In several years, such variables can be incorporated into models of teenage pregnancy. Additionally, we were unable to identify Aboriginal individuals; this is a limitation as teenage pregnancy rates are more than twice as high in the Aboriginal population than in the general population [
47]. Family and peer relationships, social norms, and cultural differences will likely never be measured through administrative data; limiting the degree to which these confounders can be controlled for.
Acknowledgements
The results and conclusions are those of the authors and no official endorsement by the Manitoba Centre for Health Policy, Manitoba Health, Active Living and Seniors, or other data providers is intended or should be inferred. Data used in this study are from the Population Health Research Data Repository housed at the Manitoba Centre for Health Policy, University of Manitoba and were derived from data provided by Manitoba Health, Active Living and Seniors and Manitoba Education under project #2013/2014-04. All data management, programming and analyses were performed using SAS® version 9.3. Aggregated Diagnosis Groups™(ADGs®) codes were created using The Johns Hopkins Adjusted Clinical Group® (ACG®) Case-Mix System” version 9.