Abstract
Keywords
Pregnancy during the teenage years is an overwhelming experience with the teenager simultaneously coping with the developmental challenges of adolescence and parenthood [1]. Children of teenage parents are likely to be disadvantaged financially and have poor health [2]. Furthermore, because young parents are less likely to complete basic schooling, they may not have the necessary knowledge to educate their children about general issues [2]. Their children are therefore at increased risk of developmental delay, poor performance in school and learning problems [3], [4]. The ongoing cycle of poverty, unemployment, poor academic achievement and physical abuse suggests that intergenerational family pathology may be a leading factor in the aetiology of teenage pregnancy.
While many studies on teenage pregnancy have explored the demographic and risk behaviours of teenage mothers, less data have evaluated the early interpersonal family relationships experienced by teenage mothers when they themselves were children [4–6]. The lack of data has led to the focus on demographic variables in addressing strategies to reduce the risk of teenage pregnancy. However, not all teenagers in ‘at-risk’ demographic categories become teenage parents, so it is clear that the story is more complex. Other factors suggested as key associations include exposure to family violence, early parental divorce or separation, adverse relationships with parents, factors relating to poor self esteem and lack of alternative life pathways and undiagnosed depressive symptomatology. The relative impact of these factors compared with demographic variables has not been explored.
We planned to explore the relative impact of demographic, early interpersonal family relationships and depressive symptomatology as associations for early age of onset of motherhood.
Method
A prospective cross-sectional cohort study was undertaken at the Royal Women's Hospital, Victoria, Australia, a tertiary referral centre for obstetrics. Institutional ethics committee approval and individual informed consent were obtained. Data from teenage (teenage) and nonteenage (control) subgroups of mothers were compared.
Subjects in the teenage subgroup of the study (teenage) were consecutively enrolled from ‘young mothers’ and general antenatal clinics. Female subjects eligible for enrolment into the teenage subgroup were aged less than 20 years. Subjects in the control group (control) were consecutively enrolled from general antenatal clinics. Female subjects eligible for enrolment into the older control cohort were all aged over 20 years. Teenage and control women were eligible for participation in the study if they did not intend to relinquish their infant, were nulliparous and had no known fetal anomaly in the pregnancy.
Assessment tools
Subjects were interviewed by a research staff member and completed several questionnaires. The first questionnaire asked about demographic variables such as the subject's age, gravidity, smoking, alcohol and illicit drug use patterns before and during the pregnancy, ethnic background, level of education and family income. Subjects also provided feedback regarding their personal assessment of their support levels by selecting an answer from a series of options relating to their social supports, housing security, emotional impression of the pregnancy and future career plans.
The second questionnaire asked questions about the subject's early interpersonal family relationships with and between their parents. The first part of the family relationship questionnaire asked about their relationship with their mother. An example question was: ‘Please take a moment to think about your relationship with your mother (or stepmother, mother figure) while you were growing up (that is, while you were a child). Please tick all the responses that apply.’ The possible responses to the question included: loving and affectionate; critical and rejecting; strict and demanding; respectful and accepting; attentive and caring; unresponsive and disinterested; understanding and sympathetic; intrusive and overprotective; or, there was no woman like this in my life. According to the items selected, responses were coded as negative, positive, mixed or absent. A negative response was coded if subjects only indicated options from the following items: critical and rejecting; strict and demanding; unresponsive and disinterested; or intrusive and overprotective. Alternatively, a positive response was coded only if subjects indicated options from the following items: loving and affectionate; respectful and accepting; attentive and caring; or understanding and sympathetic. A combination of the above positive and negative responses was coded as a mixed response. Finally, subjects who indicated that ‘there was no woman like this in my life’ were coded as absent.
The second and third parts of the family relationship questionnaire asked about their relationship with their father, and between their parents. Data were collected and coded in a similar manner. The selection of the ‘violent’ option to describe the relationship between parents was considered separately as family violence. These responses were explored by discussing the types of violence, age exposed and perpetrators in formal social work appraisals. Finally, subjects were asked to indicate whether they had experienced parental separation or divorce during their childhood.
Finally, women completed the Hospital Anxiety and Depression Scale (HADS) [7] and the General Health Questionnaire-28 (GHQ-28) [8].
In 20% of subjects, answers in questionnaires were validated by cross-referencing against data provided in an interview by a qualified social worker. From the total of 575 variables from the 20 subjects directly compared, there was agreement in 564 (agreement 98%, κ-test ratio 0.96). Thus, interobserver agreement was extremely high.
Primary outcome and power calculation
The primary hypothesis was that women, in the setting of teenage pregnancy, would be significantly more likely to describe their relationships with and between their own parents in a negative or absent fashion, compared to women over 20 years of age (controls). It was a priori estimated that 50 women in each arm of the study would have greater than 80% power to detect as significant at a two-sided 〈-error of 0.05, a difference in the incidence of negative or absent parental descriptions in the teenage and control subgroups of 35% and 10%, respectively. Therefore, the study required 100 women.
Secondary outcomes
There were several secondary outcomes. They were that women, in the setting of teenage pregnancy compared to older pregnant women, would be more likely to describe the relationships between their parents as: violent; more likely to have experienced parental separation or divorce; have significant differences in lifestyle factors such as drug use, and social factors such as education, income, housing and support; and have higher levels of psychiatric contact and psychological symptomatology, the latter as measured using HADS and GHQ-28.
Statistical analysis
Data were analyzed using SAS (Cary, North Carolina, 2002). In group comparisons of discrete data, the χ2 or Fischer's exact tests were applied. The χ2 statistic, degrees of freedom and p-value are presented for tables with more than 1 degree of freedom. Otherwise, the p-value, odds ratio and 95% confidence interval are presented. The Student's t-test or Mann–Whitney U-test were applied to generate p-values of continuous data with normal and skew distributions, respectively. To evaluate for independent associations with younger maternal age of childbirth, stepwise regression analyses were applied incorporating all variables significant at p = 0.1 on univariate analysis.
Results
Of 56 consecutive, eligible, pregnant teenage women approached to participate in the study (teenage group), informed consent was obtained from 50 (89% response). The principle reason for declining participation was a fear of a breach of confidentiality. Of 60 consecutive, eligible, pregnant women over 20 years approached to participate in the study (control group), informed consent was obtained from 50 (83% response). The principle reasons for declining participation were disinterest in the study and concern over confidentiality.
The demographics of the teenage and control cohorts were similar to data available for Victoria and Australia. The mean age of the teenage group was 17.5 years, compared with the mean age of 18 years for teenagers giving birth in Victoria. The mean age of the control group was 27.1 years, compared with the mean age of 27.7 years for women having their first baby in Victoria [9]. The proportion of the teenage group with a religious belief was 46%, and in teenage mothers Australia-wide it was 43%. The proportion of the controls with a religious belief was 82%, and Australia-wide it was 75% [10].
The background demographics of the two groups were significantly different. There were racial differences in the two maternal groups, with significantly more women of Caucasian and fewer of Asian background in the teenage group, compared to controls (p = 0.0015). The teenage group included four (8%) indigenous Australians compared to none in the control group. The educational background of the teenage group was significantly lower than controls (p < 0.0001) and remained so after adjusting for the influence of age (p < 0.0001). Family income was significantly lower in the teenage group compared to controls, with 50% of the teenage group receiving less than $15 000 per year, compared to only 8% of controls (p < 0.0001). Significantly fewer of the teenage group were married (8%) compared with 75% of controls. Significantly more teenage mothers had no fixed address or were living in a very unstable household compared to controls (homeless/very unstable home: teenage 32%, control 0%, p < 0.0001). Overall, there were significant differences in career objectives between teenage mothers and controls (p = 0.022). In subanalysis, teenage mothers were significantly less likely to have plans for any job or career in the future compared to controls (job – any/semiskilled/ professional: teenage 28%, control 54%, p = 0.0085). Conversely, significantly more teenage mothers identified ‘parent’ as their sole career choice compared to controls (teenage 42%, control 24%; p = 0.01). Teenage mothers were significantly less likely to state that they had a religious belief compared to controls (teenage 42%, control 82%; p = 0.0002). However, there were no significant differences between the two groups in terms of gravidity, previous miscarriage or abortion (all p < 0.05).
Table 1 summarizes drug use. At the start of pregnancy, significantly more of the teenage group smoked compared to controls. With the onset of pregnancy, 60% and 77% of cigarette-smoking teenage and control mothers ceased smoking. However, significantly more teenage mothers continued to smoke into their pregnancy than controls. Significantly more teenage mothers used alcohol at the start of the pregnancy compared to controls. With the onset of pregnancy, 73% and 86% of alcohol drinking teenage and control mothers quit. There were no significant differences in the numbers continuing to drink alcohol throughout pregnancy between the two groups (p = 0.09). The majority of mothers cited pregnancy as the main reason for ceasing smoking and alcohol, respectively. Significantly more teenage mothers were illicit drug users at the start of the pregnancy than controls. With the onset of pregnancy, 75% and 100% of illicit drug-using teenage and control mothers stopped using drugs. However, significantly more teenage mothers continued to abuse drugs into their pregnancy compared to controls. The spectrum of drugs used was similar between the two groups and was dominated by marijuana and amphetamines.
Drug use of the study cohort
There were marked differences in social supports between the two groups. Pregnancy was significantly more likely to have occurred in an unplanned setting in teenage mothers compared to controls (unplanned: teenage 64%, control 30%, p = 0.0003). Significantly more teenage mothers idealized the pregnancy and regarded it as the ‘single most exciting and positive event to have occurred in my life’ compared to controls (teenage 52%, control 32%, p = 0.05). Social support was significantly lower among teenage mothers than controls. Fifteen percent of teenage mothers could only rely on themselves for support with the baby, and a further 15% had only themselves and their partner. In contrast, all control mothers could identify at least one other person apart from themselves and the father of the baby, as being able to provide support about the birth (p < 0.0001). The majority of control mothers could rely on three or more other people to help them with their newborn (three or more other people available for support: teenage 40%, control 84%, p < 0.0001). A previous psychiatric history was significantly more common in the teenage group than controls (teenage 18%, control 0%, p = 0.002).
Table 2 summarizes the childhood familial relationships of the female cohort. Significantly more teenage mothers reported a negative or absent relationship with their mother and/or father compared to older controls. Of note, in the teenage group, significantly more maternal, as compared to paternal, relationships were described in a positive manner (p < 0.0001). However, there was no significant difference in the proportion of positive maternal and paternal relationships in controls (p = 0.21). Teenage mothers reported more negative and absent relationships between their parents than controls. Correspondingly, they also reported significantly fewer positive parental relationships than controls. Childhood exposure to a violent parental relationship was significantly more common in the teenage group, as compared to controls. Childhood experience of parental separation and/or divorce under 5 years of age was also significantly more common in the teenage group, as compared to controls.
Childhood familial relationships of study cohort
Table 3 summarizes the HADS and GHQ-28 data of the study cohort. The total HADS score and both subscale scores were significantly higher in teenage, as compared to control women. The GHQ-28 total and anxiety subscale scores were also significantly higher in teenage, as compared to control women.
HADS and GHQ-28 scores of study cohort
In order to determine the variables that might have an independent association with teenage pregnancy, stepwise multivariate analysis was performed on all factors significant on univariate analysis, with the p-value for inclusion set at 0.1 or less. The model included race, level of education, family income, housing status, career plans, religion, smoking, alcohol, illicit drug use, pregnancy planning, idealization of pregnancy, social supports, psychiatric history, childhood relationship with the mother, father and between the parents, childhood exposure to parental violence and parental separation/divorce under the age of 5 years, HADS-A and HADS-D subscale scores. The dependent variable was age at the time of delivery of the newborn. Childhood exposure to parental separation/divorce and domestic violence had the greatest strength of association with early age of onset of pregnancy. Other factors significant on multivariate analysis were illicit drug use, idealization of pregnancy, low family income, HADS-A and HADS-D subscale scores greater than 8 and low level of education. These data are summarized in Table 4.
Significance of patient factors on multivariate analysis
Discussion
Although demographic issues surrounding teenage mothers have been well researched, relatively little is known about the quality of the early life-relationships experienced by teenage mothers. This is the first Australian study to model the multiple risk factors identified in univariate analyses of teenage pregnancy derived from epidemiological and cohort studies against in-depth interview data that covered demographic, social, family and emotional domains. We observed that the strongest associations with a younger age of motherhood arose from adverse early life experiences in the teenager's home. The factors with strongest independent association with teenage motherhood were childhood exposure to parental separation or divorce and childhood exposure to family violence. Other independent associations were seen in respect to illicit drug use, idealization of pregnancy, low family income, the presence of psychological symptomatology in the domains of both anxiety and depression, and a low level of education.
Teenage motherhood was positively correlated with a childhood experience of parental separation/divorce. The finding is supported by retrospective and epidemiological data from other countries. In a UK study, women who were raised in single parent families due to parental separation were at increased risk of becoming teenage parents [3]. A retrospective epidemiological study from the USA analyzed birth data for the years 1995–1996 and reported a strong positive relationship between single-parent households and teenage pregnancy [11].
While the prevalence of a personal history of physical, sexual and/or emotional abuse in teenage mothers has been reported to be higher compared with older childbearing women [12], the relative strength of association of early childhood exposure to family violence and age of onset of motherhood has not been well explored. We found that a childhood experience of violent parental relationship was an independent association of younger age of motherhood. The finding is consistent with a small Israeli study, where teenage mothers were found to have a higher chance of experiencing childhood family violence than their non-pregnant peers [13]. Witnessing a violent relationship between one's parents creates an extremely hostile childhood environment, where the child may be scared of the abusing parent/s, communication may be poor and the parent–child relationship compromised. Parental separation may also occur, resulting in an unstable family structure. Having witnessed the continual battering of a parent, the child may leave home. This may then bring about a range of consequences, such as homelessness, educational and financial disadvantage, lack of social support and a higher chance of committing health-risking behaviour due to an absence of parental guidance. There may be an ongoing intergenerational cycle of abuse.
Illicit drug use was significantly associated with younger age of onset of motherhood. This association has been previously reported, with one US study estimating that illicit drug use increased the risk of an adolescent's involvement in teenage pregnancy by fourfold, compared to those with no history of substance abuse [14]. However, in some cases, teenage pregnancy may reduce drug use. In a cohort study of 456 pregnant teenagers, 66% of drug-using teenagers quit when they became pregnant [15]. In the present study, 75% of illicit drug-using teenagers quit immediately before or during early pregnancy.
Idealization of pregnancy emerged as a theme in our study. The majority of the teenage subgroup regarded their pregnancies as the single most exciting and positive event to have occurred in their lives. The finding is consistent with previous Australian research, where the pregnancy was reported to be an event the teenage mother had planned and desired. A child was believed to provide unconditional love [16]. In one US study, teenagers stated that the pregnancy was a way to reconnect with family members and it was regarded with high expectations [4].
Low family income was a significant independent association of teenage motherhood. The finding is consistent with overseas research, where the association between socioeconomic deprivation and teenage pregnancy is strong. In a large epidemiological study involving researchers from five developed countries, researchers reported that having a low family income increased a young woman's chance of early parenting by tenfold [17].
We observed high levels of psychological symptomatology in the teenage mothers compared to older antenatal women. Data from the older antenatal women were consistent with data in help-seeking Australians obtained in the 1997 Australian National Survey of Mental Health and Well-Being [18]. However, levels of psychological symptomatology in teenage mothers were significantly higher with approximately 50% and 25% of subjects meeting the cut-off score of 7/8 for a possible case in the HADS-A and HADS-D subscales, respectively, and 75% meeting the 6/7 cut off on the GHQ-28. We have previously reported high rates of teenage antenatal patients meeting one of the DSM-IV diagnoses for a mental illness [19]. Interestingly, the HADS, but not GHQ data remained significantly associated with age of onset of motherhood in multivariate analysis. Of note, despite the high levels of psychological symptomatology, few mothers had received formal contact with psychiatric or psychology services.
In the present study, poor educational attainment was a significant independent association of teenage motherhood. The finding is supported by a US study, where teenage mothers were found to complete 1.9–2.2 fewer years of education compared with women who delayed childbearing [20]. A second US study concluded that having a teenage birth would result in a 50% reduction in the likelihood of high school completion compared to non-pregnant peers [21]. The lack of educational opportunity may result in a lack of career ambition and motivation. In this study more teenage mothers were unsure of their career objective and identified parenting as their only future role. Perhaps it is not surprising that at a time when their educated peers begin studying for their ‘working careers’, those with ‘mother’ as a career choice fall pregnant.
The study has a number of limitations. Participants were recruited from a single institution which may impose constraints on generalizability. Although events enquired into are highly likely to be accurately recalled, the study includes retrospective data and the potential for recall bias is present and this could influence findings.
While school-based educational programs and availability of contraceptives are important to help reduce teenage pregnancy, it is likely that strategies to lower rates of teenage births will require more complex and multilevel interventions. These should start by identifying vulnerable new parents and offering them extended home-visiting services, parenting support and family violence interventions to help improve the environment into which they rear the next generation, especially over those important early 5 years. Strategies will then need to focus on self-esteem and reducing the idealization of pregnancy as a solution to general life dissatisfaction, with strategies to entice school students diverging from educational pathways into alternative, positive career choices that do not rely on formal education skills. Finally, adverse psychological symptomatology in teenagers needs to be identified at every opportunity.