Background
Teenage pregnancy is a significant public health issue. Giving birth during adolescence is strongly associated with adverse living conditions in later life [
1]. Approximately 1.25 million teenagers become pregnant each year in the 28 OECD (Organisation for Economic Co-operation and Development) nations [
2]. Of those, about half a million pregnancies will be terminated and approximately three quarters of a million teenagers will become mothers. In 2003, The Netherlands, Sweden, Denmark, Finland and Slovenia had the lowest adolescent birth rates in Europe (6/1,000) while the United Kingdom (27/1,000) was characterized by the highest rates [
3].
In Germany, the proportion of mothers between 10 and 18 years of age rose from 0.9% in 2000 to 1.0% in 2006 (19/1,000 women of the same age) [
4]. Among women who had an induced abortion the proportion of adolescent women rose from 4.7% in 2000 to 5.5% in 2006 (17/1,000 to 19/1,000). In the western federal state Lower Saxony between 2000 and 2006 the rate of live births ranged from 22 to 18 while the rates in Berlin and in the eastern federal states were 20–70% higher during this period of time. The same regional pattern can be observed regarding the rates of induced abortions [
4].
Recently, attention has been focused on subsequent pregnancies among teenage mothers. The likelihood of a second birth among adolescent mothers is much greater than the likelihood of a first birth among teen females who have not had a child yet. For example, in the United States in 2001, there were 35.7 births per 1000 females aged 15 to 19 years who never had a birth compared to 175.1 births per 1000 females aged 15 to 19 years who previously had one birth [
5]. Accordingly, twenty percent of teen births occurred to young women who had been mothers already. In a representative sample of adolescent mothers in the US about two thirds reported that the second pregnancy was not intended [
6]. In Germany, among those teenagers who had an abortion in 2006, 2.8% reported to have a child already [
4].
A small number of studies have examined the relationship between parity and reproductive outcomes among teenagers. Cross-sectional studies suggest a lower risk for low birthweight [
7] and for neonatal, postneonatal and infant mortality [
7‐
9] in the first pregnancy of adolescent women. However, the results of longitudinal studies are similar to studies on adult women populations [
10], and indicate that higher rates of low birthweight infants [
11] and intrauterine growth retardation [
12] are associated with teenagers' first birth compared to their second.
Previous studies on parity and reproductive outcomes in adolescents had several important limitations. Cross-sectional studies based on the linking of birth and death records could not access relevant confounders [
7,
9]. Longitudinal studies that followed individuals over time had samples that didn't reflect the general population, were based on small geographical areas and/or lacked statistical power because of a small sample size [
12‐
15]. Only one longitudinal study considered the adolescents' obstetric history regarding abortions and miscarriages [
13]. The objective of this study was to compare the perinatal outcomes (rates of stillbirths, neonatal and perinatal mortality, preterm birth and very low birthweight) of nulliparous teenagers and teenagers who previously had an induced abortion or a live- or stillbirth after adjustment for potential confounders (maternal nationality, partner status, smoking, prenatal care and pre-pregnancy BMI).
Discussion
Using routinely collected perinatal data we examined the relationships between obstetric history and reproductive outcomes among adolescents. Compared to nulliparae, adolescents with a previous birth had a more than twofold higher risk for perinatal mortality and a more than fourfold higher risk for neonatal mortality.
The results of this study confirm the findings of studies with cross-sectional designs. For example, Hellerstedt et al. [
8] found that the crude risk for neonatal deaths was 20 percent higher among multiparae compared to primiparous teenagers. Additionally, in the US, the risk for neonatal mortality was about 1.5-fold increased among 18–19 year old multiparae compared to primiparae of the same age [
7]. Our study contradicts findings from a longitudinal US-study that found an almost twofold higher risk for perinatal deaths and a threefold higher risk for stillbirths among nulliparous women compared to adolescents with a previous birth [
13]. However, the study was underpowered because of a small sample size and the results were not statistically significant. Other longitudinal studies also had limited sample sizes and could not examine rare outcomes such as perinatal mortality [
11,
12,
14].
Due to the lack of studies on adolescent multiparae that considered confounders, the selection of potential confounders in this study derived from research on adolescent primiparae or adult women [
18,
20,
21]. The confounders we examined did not explain the elevated risks for perinatal and neonatal mortality among adolescents with a previous birth compared to nulliparous teenagers. Rather, the adjustment for confounders strengthened the observed relationships. To inform future preventive efforts, further studies should attempt to identify the mediating factors that increase the risk among adolescents with a previous birth for neonatal and perinatal mortality.
Previous studies on the risks associated with a history of abortion among teenagers are sparse. Lao and Ho [
22] found that a previous induced abortion among Hong Kong teenagers was not related to a higher risk for preterm birth. In our study, teenagers with a history of an abortion had a 3.4-fold higher risk for a stillbirth and a 2.2-fold higher risk for a preterm born infant than nulliparous adolescents. After adjustment for confounders these associations disappeared. In our study and in Hong Kong [
22] teenagers with a previous abortion were characterized by a much higher smoking rate than the adolescent mothers with no abortion history. In the same group of teenagers, compared to nulliparous women, the risk for a very low birthweight infant was increased in the adjusted model. Because the confounders we examined (such as smoking during pregnancy or inadequate prenatal care) are related to both stillbirths [
23] and very low birthweight [
24] further studies are needed to understand this contradictory result.
In our sample, the rates of adverse outcomes largely correspond with Scottish data [
10] but they were lower compared to the American studies of Blankson et al. (1993) [
12] and Hellerstedt et al. (1995) [
8], especially regarding preterm birth. However, neither the ethnic composition nor the social context of these US studies and our study can readily be compared. One reason for the lower incidence rates in preterm birth in Lower Saxony may be sought in the fact that the perinatal registry does not cover (planned and unplanned) out-of-hospital births. Also, the incidence rate of adolescents' pregnancies in Lower Saxony is slightly below the German average. This may reflect a less adverse environment compared to those areas with higher incidence rates such as Berlin and the eastern federal states, areas with higher unemployment rate, in particular among adolescents.
We have no information on known risk factors for adverse outcomes, especially of teenage pregnancies, such as domestic violence, stress, or poverty [
20,
25]. Smith and Pell compared the birth outcomes of primiparae and secundiparae between adolescent and adult mothers and found no differences among primiparae but higher perinatal mortality among adolescent than among adult secundiparae [
10]. Smith and Pell concluded that a second birth during adolescence probably occurs more often in the context of poverty and poor nutrition than a second birth among mature women. Therefore, among teenagers living in a disadvantaged social context the accumulative burden of a second pregnancy may result in adverse outcomes. Our sample, however, is characterized by a vast proportion of migrant adolescents who usually are married and have access to strong support networks within their communities. The differences regarding the social context and behavioural characteristics (such as migrant status, lone motherhood, and smoking) among the three reproductive groups in our sample may point at the necessity for sociodemographically tailored approaches when attempting to improve the reproductive health of these women.
According to Klerman, findings from previous cross-sectional studies may be biased because they often compared any higher order births to nulliparous teenagers and missed important confounders [
26]. The results of our study cannot readily be compared to these studies because we deleted adolescents who previously had a spontaneous miscarriage from the sample and thus only examined "true" nulliparae. Contrary to Hellerstedt et al. we compared the outcomes of the first birth to the outcomes of the second pregnancy while higher order pregnancies were excluded [
8]. Additionally, we considered the outcome very low birthweight (< 1500 grams) instead of low birthweight (< 2500 grams) because the predictive value of the latter variable for children's health is still being debated [
27]. We did not use intrauterine growth retardation as an outcome because the underlying growth norms refer to the German population and may not be valid for migrant newborns who account for more than one third in our sample. Previous studies on subsequent teenage pregnancy mostly have been from the US. However, the German adolescent population differs from the US population on several important aspects such as the ethnic composition. Also, contrary to the US, in Germany, prenatal care is free for all women regardless of their age, migration or employment status. Health insurance is mandatory. For refugees and for unemployed women the costs of prenatal care are covered by the community. Consistent with other studies from countries with free provision of prenatal care a huge proportion of teenagers in each reproductive group chose not to utilize this offer [
28]. These adolescents may be characterized by a lower level of knowledge about the availability of prenatal care. Those who have had a previous birth or an abortion may anticipate negative comments on their condition by health care providers [
28].
Our study has several limitations. To suggest causality this type of study has to be longitudinal. Adolescents who have a birth following a prior birth or an abortion are different in many ways from those who have a first birth with no previous pregnancies [
26]. We have controlled for nationality, partner status, smoking, pre-pregnancy BMI and adequacy of prenatal care to differentiate between adolescents with different reproductive histories. However, we have no information on birth spacing. It may well be that a short inter-pregnancy interval is one of the underlying causes of worse outcomes among adolescents who had a previous pregnancy [
29]. Intimate partner violence is another known risk factor for subsequent pregnancies during adolescence that we were not able to examine [
30]. Alcohol is a known teratogenic substance that operates under a dose-response mechanism and drug use is associated with adverse pregnancy outcomes as well [
31,
32]. We could not access information on these substances. Further known risk factors for adverse pregnancy outcomes such as an unwanted pregnancy, stress, poverty, and vaginal infections also are not assessed in the routine perinatal survey. In summary, it is possible that our findings might be eliminated if we had accessed more confounders or if the study had a longitudinal design.
Although the rate of non-participating hospitals is rather small (2–13 percent), we cannot rule out a selection bias. Small numbers in some cells resulted in broad confidence intervals.
Induced abortions usually are recorded in the mother's passport but in the next pregnancy the women can choose to visit a new gynaecologist or midwife without bringing her mother's passport and thus deny the previous pregnancy. Therefore, underreporting of previous abortions is possible may have resulted in a classification bias.
A high proportion of teenagers who already gave birth to a child were characterized by Non-German nationality. Although we adjusted for nationality we cannot disregard that residual confounding may have occurred and that characteristics associated with migrant status might have affected the risk for adverse outcomes.
However, the current study expands on previous studies in several ways: it is not based on vital statistics but rather on data from a population sample prospectively collected by physicians and midwives. Thus, external validity and generalizability are satisfactory. Additionally, we incorporated several relevant confounders such as smoking in our analyses. Our study not only considers the adolescents' parity but draws attention explicitly to the reproductive history.
Competing interests
The author(s) declare that they have no competing interests.
Authors' contributions
BR analysed the data and mainly wrote the manuscript. BS assisted with interpreting the results and writing the manuscript. PW assisted with interpreting the results and writing the manuscript. The manuscript has been read and approved by all three authors.