Intervention Now to Eliminate Repeat Unintended Pregnancy in Teenagers (INTERUPT): a systematic review of intervention effectiveness and cost-effectiveness, and qualitative and realist synthesis of implementation factors and user engagement

Social–cognitive–ecological theory

Social–cognitive–ecological theory31 addresses cultural norms, modelling and the concepts of self-efficacy and support. It is based on Bandura’s32 postulation that a person’s internalised standards for behaviour are developed from information conveyed by sources of social influence (parents, peers and characters portrayed in mass media) and their environment. For example, a young mother may desire a repeat pregnancy to seek a sense of fulfilment as a woman, to create a stable relationship with her partner or to improve connections within her family.

Developmental theories

Erikson33 posited that an adolescent has physical, cognitive, socioemotional and moral goals to accomplish during their transition to adulthood. Developmental theories described in repeat pregnancy literature address this transition. In addition, they address the management and perception of pregnancy by adolescents, and the relationship between risky sexual behaviour and pregnancy. These theories also examine the development of health promotion behaviours in families led by adolescent parents. 34

Resilience theory

The concept of resilience can be described as ‘the process of identifying or developing resources and strengths to flexibly manage stressors to gain a positive outcome, a sense of confidence or mastery, self-transcendence, and self-esteem’. 35 Adolescent resilience theory focuses particularly on the ‘assets and resources that enable some adolescents to overcome the negative effects of risk exposure’. 36

Recoil–rebound theory

The main focus of this theory is the parenting adolescent mother’s resilience, social influences, recoil (having a repeat adolescent pregnancy) and rebound (the use of contraception or abstinence to prevent a repeat adolescent pregnancy). 37

Resilience–recoil–rebound theory

This model, proposed by Porter and Holness,37,38 has four central concepts: the adolescent, pregnancy, recoil–rebound interactions and resilience. The external influences of family, peers, school, church and community, though incorporated, are not major concepts within this theory.

Databases searched

We conducted initial scoping searches to identify the volume and nature of the literature we were likely to encounter. Building on and revising those search strategies, we searched the major, relevant electronic databases (listed in Box 1) for published literature using strategies that combined thesaurus terms and keywords relating to pregnancy, termination of pregnancy (TOP) or parenthood with ‘adolescence’ and text word synonyms for ‘repeat’ or ‘subsequent’. These searches were conducted between May and July 2013 and updated in June 2014.

Design of search strategies

We used a search strategy developed and piloted in MEDLINE and subsequently modified for use in the other databases (full details of all the search strategies can be found in Appendix 1). Further modified versions of the same strategy were used to search the following databases for ‘grey’ literature: OpenGrey, Scopus, Scirus, Social Care Online, National Research Register, NIHR Clinical Research Network Portfolio and Index to THESES. To capture economic studies, we searched Research Papers in Economics (RePEc) and the American Economic Association’s electronic bibliography (EconLit). We applied an alternative search strategy specifically designed to capture the type of descriptive titles that are common in qualitative studies to the following selected databases: Sociological Abstracts (SocAbs), Applied Social Sciences Index and Abstracts (ASSIA), British Nursing Index (BNI) and the Social Science Citation Index (SSCI). This strategy combined additional synonyms for ‘pregnancy’, ‘adolescence’ and ‘repeat’ with a brief qualitative filter comprising three broad free-text terms, ‘qualitative’, ‘findings’ and ‘interview$’, that has been shown to be as effective as a more complex one. 42,43

A number of systematic reviews related to teenage pregnancy have been published in the last two decades. The earliest we found in our scoping search was dated 1997. 44 Therefore, after allowing for an additional conservative margin, we limited our searches to 1995 onwards. We intended to conduct a separate sensitive search excluding the terms for second or subsequent pregnancies but including a filter for systematic reviews as a means of capturing relevant data from earlier studies; however, our expert panel advised us that social norms and behaviours relating to teenage motherhood have changed dramatically and that they considered evidence related to contraception more than 10 years old to be of limited relevance. Therefore, we omitted this additional search.

Owing to the dearth of UK literature identified in the database searches, we developed an additional strategy to identify UK grey literature using the search engine Google (Google Inc., Mountain View, CA, USA). This strategy comprised 24 search terms that were variations on ‘repeat’ or ‘second’ and ‘teenage’, ‘teenager’, ‘teenagers’, etc. and ‘pregnancy’ or ‘pregnancies’, each prefaced by the term ‘site: UK’. This search was conducted in January 2014.

In July 2014, we conducted repeat searches by (1) rerunning or updating the searches from all the relevant databases; (2) citation tracking the RCTs and systematic reviews identified during the course of our literature searches using ‘forward chaining’ and ‘backward chaining’ techniques;45 and (3) using the Google Scholar alert function to indicate any new trials on repeat adolescent pregnancy.

The issues surrounding implementation of interventions are a primary focus of this systematic review; qualitative and process evaluation evidence associated with trials is highly relevant, particularly to better understand the facilitators of and barriers to implementation. 46 Therefore, a key feature of our search strategy was to identify ‘evidence clusters’, that is studies that investigated the implementation or acceptability of interventions related to key RCTs. 47 These studies have the potential to indicate the effectiveness of an intervention as well as its acceptability to users and barriers to its implementation and uptake. However, we were unable to capitalise on the full potential of the approach as we wished to explore UK-relevant implementation issues, and all but one of the trials we identified were conducted in the USA and the other was conducted in Australia.

Finally, also in July 2014, we compiled a list of journals that had published the RCTs and the qualitative studies included in our review and compared this list with the Master List of hand-searched journals and conference proceedings (journals/conference proceedings being hand-searched by Cochrane Entities);48 we hand-searched the online indexes of 17 English-language journals that did not appear in this Master List from January 2010 (see Appendix 1). Because of time constraints and practical difficulties we did not search the index of one foreign language journal.

References were managed using EndNote bibliographic reference management software (Thomson Reuters, CA, USA) and extracted data were organised in a database using Microsoft Access^® (Microsoft Corporation, Redmond, WA, USA). Two reviewers independently screened titles and abstracts to identify potentially relevant documents, which were retrieved and assessed according to the inclusion criteria described below (see Study inclusion criteria). Disagreements were resolved by discussion or, when necessary, by a third reviewer.

Mapping exercise

In November 2013, we facilitated a meeting with a group of stakeholders, including public health, primary care, sexual health, obstetrics and midwifery representatives, as well as the full review team. A report of the meeting, including a list of those who attended, can be found in Appendix 2. We presented initial results from our mapping exercise to describe the scope of the literature we had identified and we conducted a workshop with three discussion groups which focused on linking the different topics to the group members’ experiences. We asked the subgroups at the meeting a range of questions; these are listed below:

• How relevant is evidence from different settings and particularly from non-UK studies to your context?

• What components of complex interventions do you consider to be most important in relation to reducing repeat pregnancies?

• How relevant are other outcomes (health-related, social, educational, maternal or child-related, etc.) to repeat pregnancies?

• Are repeat teenage pregnancies truly unintended? (What are the differences between unintended, unplanned and unwanted pregnancies?)

• In your experience, what are the differences between the motivating factors for repeat teenage pregnancies and first teenage pregnancies?

• Which intervention components best address the factors that motivate young women to have (or not to avoid having) a repeat pregnancy?

• What are the barriers to and facilitators for accessing interventions?

• What is the best way of packaging and presenting an intervention to teenage mothers to ensure maximum uptake (such as pregnancy spacing, making healthy choices, using contraception effectively, planning your future or others)?

• How is the success of a programme best measured in monetary terms?

The meeting was concluded with feedback from the groups and discussion.

Presentation of preliminary findings and feedback exercise

In June 2014, a second meeting was facilitated as part of the work of the Task and Finish Group which co-ordinates the PHW response to teenage conceptions. Members of this group included public health, primary care, sexual health, obstetrics and midwifery representatives, as well as members of the review team. A report of the meeting, including an attendee list and comments made by stakeholders, can be found in Appendix 3.

We presented an overview of the study and progress to date, together with Phase 3 results from the quantitative and qualitative syntheses of the study, as well as preliminary findings from our realist synthesis, and invited feedback from the delegates.

We then presented the delegates with the following questions and received comments:

• What do these findings mean to you within your work context?

• Who do you think should hear these findings?

• How do you think these findings should be delivered to these audiences to maximise uptake?

• Is there a policy message?

• Do these findings make sense to you?

• How similar are they to the interventions you are delivering locally?

The findings from these meetings contributed to the realist synthesis (see Chapter 4) and informed the overarching synthesis (see Chapter 5).

Service user engagement

Two members of our team presented the findings of the review to a group of 17 young mothers, ranging in age from 15 to 22 years in April 2014. Flying Start helped us to organise and facilitate this meeting. Based on the evidence generated during the review, we sought opinions on the following three areas: (1) contraception, (2) psychosocial programmes and (3) barriers to and facilitators of the uptake of interventions. The findings were presented to the group through a range of person-centred activities. For example, to present the findings regarding contraception we used pictures of each contraceptive method on a flipchart with space to record summary points and keywords. The views expressed and the discussion that was stimulated are summarised in Appendix 4 and were used to inform both the realist synthesis and the narrative synthesis of this report.

Study inclusion criteria

We included published studies from any country. Our search strategy was designed to capture published studies of any type, including trials of interventions, effectiveness studies, interrupted time series studies, cost-effectiveness studies, process evaluations, surveys and qualitative studies of participants’ views and experiences of interventions. We also considered relevant grey literature of any type, such as unpublished reports, service evaluations and theses, but limited this to UK-based literature so as to be applicable to the NHS and UK public health bodies. We set the earliest date for published work as 1995 and excluded any studies that reported data collected prior to 1990. We described the inclusion criteria using the patient intervention comparison outcome (PICO) framework. 49

Study exclusion criteria

We did not exclude quantitative studies found in non-English-language publications; we used English-language abstracts, if available, to help decide whether or not they met our inclusion criteria and an online language translation service, Google Translate, to aid data extraction. Abstracts in which results could not be confirmed in subsequent publications were excluded. We found that translation software was adequate for the quantitative data extraction of baseline variables and outcome data but we were not confident of its reliability for the quality appraisal of studies; therefore, we asked a Spanish-speaking team member to appraise the quality of the foreign-language studies (which were published in Spanish and Portuguese).

A priori we had decided to exclude non-English-language qualitative studies if we had no native speakers available because of the complex nature of the linguistic validation of translations on such nuanced work. We did not, however, encounter any such papers so there was no loss of data.

We limited our searches for unpublished material to the UK to enhance the direct applicability of the results to the NHS and UK public health bodies. This judgement was supported by advice from the advisory group obtained during the stakeholder engagement exercise. Once studies found in the grey literature had met a very basic quality standard (see Quality appraisal section) they were included in the in-depth review and we incorporated further judgements about study quality when interpreting the evidence.

Prioritising evidence for the in-depth review (further study inclusion criteria)

The review team, with advice from independent experts obtained during the stakeholder engagement exercises, prioritised the evidence for in-depth review. Beginning with evidence relating to primary outcomes and within each grouped set of data, we prioritised the best quality evidence of most relevance to address the research questions. In order to systematise this process we applied the completeness, accuracy, relevance and timeliness (CART) framework, whereby evidence is judged on the criteria of completeness, accuracy, relevance and timeliness. 51 We developed a protocol for applying CART criteria which can be found in Appendix 5. We screened all the studies using these criteria.

Study classification: Effective Practice and Organisation of Care classification

We classified all the quantitative papers using the flow chart shown in Appendix 6, in accordance with the Effective Practice and Organisation of Care (EPOC) Group’s non-randomised studies methods guidance on incorporating diverse types of evidence in a review (see Appendix 19). These guidelines classify the study designs to use for the evaluation of the effects of health-care interventions. If quantitative papers did not meet inclusion criteria for the quantitative effectiveness evaluation but did provide insights for either the qualitative or realist syntheses, they were transferred to that evidence stream for assessment. All studies that contained epidemiological data, no matter how they were classified according to EPOC criteria,52 were considered as part of the risk factor identification.

Quality appraisal

The protocol for part of this review is registered with Cochrane and we intend to be publish some of the findings as a Cochrane review. Therefore, RCTs and quasi-RCTs were assessed using Cochrane’s risk of bias tool. 53 We categorised and reported the overall risk of bias for each of the included trials as having:

• a low risk of bias (plausible bias unlikely to seriously alter the results) if all criteria were met

• an unclear risk of bias (plausible bias that gives rise to some doubt about the results) if one or more criteria were deemed unclear

• a high risk of bias (plausible bias that seriously weakens confidence in the results) if one or more criteria were not met.

The Cochrane risk of bias tool was supplemented by the mixed-methods appraisal tool (MMAT)54 (see Appendix 7 for mixed studies reviews). This MMAT has the advantage of incorporating the appraisal of several different study designs (qualitative, RCT, non-RCT, observational and mixed methods) using a single tool with a coherent range of quality criteria. 55

We slightly amended the published version of the MMAT54 for the qualitative studies by adding an alternative response, ‘somewhat’, for items that were partially described if we considered that either a ‘yes’ or a ‘no’ response would be misleading. We used the MMAT assessment items relevant to the qualitative analysis as follows:

• Are the sources of qualitative data (archives, documents, informants, observations) relevant to address the research question (objective)?

• Is the process for analysing qualitative data relevant to address the research question (objective)?

• Is appropriate consideration given to how findings relate to the context, for example the setting, in which the data were collected?

• Is appropriate consideration given to how findings relate to researchers’ influence, for example through their interactions with participants?

Pilot studies suggested that the MMAT54 was an efficient and reliable tool. However, it does not cover economic evaluations; therefore, for such evaluations, we used the Drummond checklist, which is recommended for economic studies56 (see Appendix 8).

Descriptive study summaries

We present the full findings of the data extraction exercise for studies included in the meta-analysis of RCTs and in the qualitative synthesis in a table of study characteristics (see Appendix 18). These include the study details, the setting, the population, the quality appraisal and the methods. We present sociodemographic characteristics known to be important from an equity perspective. For this process, the PROGRESS (place, race, occupation, gender, religion, education, socioeconomic status and social status) framework was utilised. 59 For the studies that were included in the risk factor analysis, a brief summary table containing the study type and PROGRESS data is included (see Appendix 19).

Quantitative meta-analysis

Investigation of risk factors using metaregression

A key aim of the review was to search, identify and summarise the populations of young girls that are at greatest risk of repeat pregnancy (by, for example, considering income, social deprivation, ethnicity, degree of rurality, substance misuse, whether or not currently in care or care leavers, and those from vulnerable or at-risk communities). We expected that these factors would be considered and summarised as important confounding variables within RCTs, quasi-RCTs and cost–benefit analyses for studies investigating interventions to reduce repeat pregnancies. We also aimed to identify non-interventional studies that present epidemiological data (e.g. cohort studies, cross-sectional studies and policy documents) of effect modifiers associated with an increased risk of repeat pregnancy. We shortlisted possible risk factors from a review by Rigsby et al. 17 and added any risk factors which were supported by several studies. Data were extracted and summarised and, if possible (if 10 or more studies reported similar factors), presented in a metaregression summarising the standardised effectiveness by presenting the slope parameter, with the associated 95% CIs. Graphical representations of the metaregressions were plotted using the square root of the sample size for each of the studies.

Cost-effectiveness

We provide a narrative review of economic evaluations of interventions specifically designed to reduce the number of unintended repeat teenage conceptions. We planned to stratify any economic studies found by the public health lever mechanism used; examples of such mechanisms are government, statutory or legal mechanisms, public information mechanisms, school-based group or targeted interventions, and NHS- or charity-initiated mechanisms. We were particularly interested in the perspective of the analysis or the type of economic evaluation (e.g. cost analysis, cost–benefit analysis, cost-effectiveness analysis or cost–utility analysis). We documented the way in which these studies attempted to overcome the methodological challenges specifically associated with these types of complex, preventative, behavioural change based interventions. 63–66 We planned to conduct a meta-analysis of the economic evidence (if the data were sufficiently homogeneous to allow it).

Qualitative studies

For qualitative studies, we developed, a priori, a coding framework adapted from the Support Unit for Research Evidence (SURE) framework67 for identifying factors affecting the implementation of a policy option (see Appendix 12), which was successfully used in a qualitative systematic review of barriers and facilitators to the implementation of lay health worker programmes to improve child and maternal health. 47 Data were coded directly from the papers into the framework and thematic syntheses were conducted using a framework method developed by Ritchie and Spencer. 68

Surveys, process evaluations and other types of data

Data from surveys, process evaluations and other sources, such as reports, may be either semi-narrative or quantitative or both. Data were extracted to present evidence of acceptability and uptake of interventions, and were synthesised in a narrative summary and aggregated using thematic analysis. 69

Realist synthesis

We selected subsets of evidence and applied the principles of realist synthesis. 58,70,71 We identified explicit or implicit theories which postulate how an intervention has an underlying causal mechanism that works in a defined social context to result in a particular outcome. Such theories may also be used to explain the failure of an intervention. Additional theories were identified from the wider literature (e.g. policy documents), the advisory group members or personal contact with other experts in the field. Data synthesis involved individual reflection and team discussion in order to question the integrity of each theory, adjudicate between competing theories, consider the same theory in different settings and compare the particular theory with actual practice. 72 Coded data from the studies were then used to confirm, refute or refine the candidate theories. Thus, we attempted to explain which interventions work, for whom and under what circumstances.

Preferred Reporting Items for Systematic Reviews and Meta-Analyses diagram

The PRISMA diagram77 (Figure 3) describes the full flow of studies through the filtering process to our final 77, including seven studies that were added as a result of the updated literature searches. Excluded studies and reasons for exclusions can be found in Appendix 16.

FIGURE 3.

The PRISMA diagram illustrating the sifting of evidence from discovery to inclusion in the review.

Content or intervention type

Interventions were predominantly of two types: (1) those that focused solely on promoting the uptake, or facilitating the timely postpartum uptake, of contraception, usually long-acting reversible methods delivered by injection, implant or intra-uterine devices, or oral contraception; and (2) complex interventions comprising various combinations of health, social and educational elements, one of which was usually a contraception regimen. Complex interventions often had multiple objectives which included promoting parenting skills and infant health and nutrition, as well as preventing repeat pregnancy. Specific components that we identified included:

• Fertility health-related components:

  • pregnancy testing and maternity counselling

  • primary and preventative health-care services (including prenatal and postnatal care)

  • counselling and referral for family planning services

• Sexual health education and guidance:

  • referral for screening and treatment of sexually transmitted infections, including HIV/AIDS

  • educational services related to family life and problems associated with adolescent premarital sexual relations

• Other maternal health components:

  • nutrition information and counselling

  • mental health services

• Child health:

  • referral to appropriate paediatric care

• Family support:

  • adoption counselling and referral services

  • addressing domestic violence and peer relationships

  • supportive counselling and social work services

• General educational interventions:

  • Appropriate educational and vocational services including job skills training, tutoring and mentoring.

Overall summary of included studies and study quality

Quantitative data

Quantitative data related to potential risk factors for unintended repeat teenage conceptions were extracted from the 31 quantitative studies, as well as the RCTs, cohort studies and other, non-comparison longitudinal intervention studies if relevant and possible. The results of the metaregression showed inconsistent outcome reporting between studies that predominantly reported the repeat pregnancy proportion. Of all of the outcomes, only adequate data were provided to enable consideration of the factors that determine repeat pregnancy. Likewise, data were sparse for many other risk factors of interest and we were only able to evaluate the risks if at least 10 studies provided data. We extracted all the relevant risks, if data were available, if two or more papers included that risk factor. The risks we evaluated were the proportion of young people who:

• had a minimum education

• had been abused

• smoked

• lived with a partner (father of their children)

• used LARCs

• used oral contraception

• used other forms of contraception.

We extracted data on other risk factors but had insufficient data to perform metaregression with any statistical integrity. All the studies and their study types included in this metaregression are listed in Appendix 14. The risk factors we extracted but had insufficient data to assess were:

• teenage mothers’ age (mean and SD) at birth of first child

• proportion of teenage mothers receiving support from friends and family

• proportion of teenage mothers with reported depression or anxiety

• proportion of teenage mothers misusing drugs or alcohol

• proportion with partner reported as ‘older’ – (notionally a partner that is older by at least 3 years)

• age (mean and SD) at first sexual intercourse in years

• child mortality or loss related to previous pregnancy caused by miscarriage, stillbirth, abortion or child being taken into care

• maternal morbidity related to pregnancy, abortion or childbirth (e.g. hypertension, haemorrhage, pulmonary embolism, pre-eclampsia or pelvic bone size)

• proportion of teenage mothers reported to have had at least one sexually transmitted infection

• proportion of teenage mothers whose first pregnancy was desired or intended

• proportion of teenage mothers who failed to comply with a contraception regimen.

Despite the requirement for at least 10 studies to contain usable data for inclusion in the regressions, the 95% CIs around our estimates were very wide, indicating a low precision of the estimate. The results of these metaregressions cannot rule out that any of the experience factors may genuinely determine the risk of repeat pregnancy (Figures 4–10). However, from the data that we have found, there is little evidence to suggest that any of the seven risks we evaluated were helpful in determining the likelihood of repeat pregnancy.

FIGURE 4.

Prediction of repeat pregnancy (%) by continuation of education (%). The studies are plotted by the square root of their sample size. Slope parameter: –0.85, 95% CI –21.7 to 20.0; p-value = 0.93.

FIGURE 5.

Prediction of repeat pregnancy (%) by the level of abuse (%). The studies are plotted by the square root of their sample size. Slope parameter: –0.74, 95% CI –121.3 to 119.8; p-value = 0.99.

FIGURE 6.

Prediction of repeat pregnancy (%) by the level of smoking (%). The studies are plotted by the square root of their sample size. Slope parameter: 0.07, 95% CI –75.3 to 75.4; p-value = 1.00.

FIGURE 7.

Prediction of repeat pregnancy (%) by the proportion of young women that were living with their partner (%). The studies are plotted by the square root of their sample size. Slope parameter: 1.11, 95% CI –17.8 to 20.0; p-value = 0.90.

FIGURE 8.

Prediction of repeat pregnancy (%) by the proportion of young women and girls that were using long-acting reversible contraception (%). The studies are plotted by the square root of their sample size. Slope parameter: –0.74, 95% CI –27.0 to 25.3; p-value = 0.95.

FIGURE 9.

Prediction of repeat pregnancy (%) by the proportion girls that were using oral contraception (%). The studies are plotted by the square root of their sample size. Slope parameter: 1.38, 95% CI –44.0 to 46.7; p-value = 0.94.

FIGURE 10.

Prediction of repeat pregnancy (%) by the proportion of young women and girls that were using other contraception methods (%). The studies are plotted by the square root of their sample size. Slope parameter: 0.42, 95% CI –36.0 to 36.9; p-value = 0.98.

In Figures 4–10, we present the slope parameter of the linear regression; this indicates the increase (or decrease) in the percentage of young women and girls with repeat unintended pregnancy for each per cent increase in girls presenting a particular risk factor. A positive (or negative) value of the slope indicates an increase (or decrease) in the percentage of young women and girls with repeat unintended pregnancy. The slopes are presented as mean differences to highlight that we actually know very little about the actual percentage of increase (or decrease). That is why we also present the 95% CI, to highlight the level of uncertainty around the slope, and the p-value, to represent the statistical significance. A low p-value (p < 0.05) would indicate that a risk factor is likely to be a meaningful addition to our model because changes in the risk factor’s value are significantly related to changes in the repeat pregnancy rate.

Using Figure 4 as an example, the slope parameter of –0.85 can be interpreted to indicate that the percentage of teenagers with repeat unintended pregnancy reduces by 0.85% for every 1% increase in the proportion of young women and girls remaining in education. However, the 95% CI ranges from –21.7 (indicating a 21.7% decrease in pregnancy rate per 1% increase in the proportion in education) to +20.0 (indicating a 20% increase in the pregnancy rate per 1% increase in the proportion in education). These data result in a p-value of 0.93, which indicates that the mean decrease of 0.85% per 1% increase in those in education is meaningless. This interpretation also applies to Figures 5–10, taking into consideration the positive or negative values of the slope parameters accordingly.

As no quantitative evidence has emerged to help characterise the population at most risk of repeat unintended teenage conceptions, we turned to an examination of the in-depth qualitative studies which asked young women about their circumstances and motivations. Five of the 10 included qualitative studies shed light on the characteristics and life circumstances of young women, and some of the factors that help to explain why they had, or chose not to have, a repeat pregnancy. 85,129,130,134,136 Themes included contextual, motivational and emotional factors, and rationales for having a rapid repeat pregnancy.

Contextual factors: family, peer group and lifestyle

Many young mothers had unstable backgrounds, characterised by a lack of family support, insecure housing and chaotic lifestyles, and were lacking educational and vocational opportunities. Sex education and access to contraception alone could not effectively address the issue of unprotected sex in the context of these teenagers’ real lives, and they continued in their risky behaviours. 129 Although they understood the risks of unprotected sex, there was a complete lack of planning or consideration of the consequences; rather, teenage sex was a spontaneous activity132,133 and, in the context of some families and peer groups, teenage pregnancy was considered the norm. 129

You take chances, it’s just like gambling, if you keep doing it . . . I knew [about birth control]; I just wasn’t careful.

Teenage mother who received state food aid (USA)129

We’ve accepted [teen pregnancy and early childbearing]. Now it has become normal to become a parent while you’re a teenager, but I think it is time to take a role that this is not right.

Mother of teenage parent who received state food aid (USA)129

Motivational factors: goals and aspirations

A key feature of the lifestyles of young women whose environment offered little opportunity was a lack of goals and aspirations, and, in some cases, even a lack of expectation of any future achievement that could be planned for; they lived in the moment. However, some young mothers did have educational or career objectives, although there were significant barriers to achieving them. These barriers included the difficulty of combining the role of being a mother with that of being a student, particularly when children were ill, and having limited access to childcare. 129 Poorly paid work provided only a marginal improvement in income compared with state benefits for young mothers in the UK. This had to be weighed against the disadvantage of being out at work for several hours a day and thus being unable to spend much time with their children. 130 Nevertheless, those who had goals and aspirations were more likely not to have a rapid repeat pregnancy. 134 Some of the respondents recognised and reacted against the prevailing stereotypical views of teenage mothers (thoughtless, feckless, welfare-dependent, etc.) and were determined not to conform to that image. This was a powerful motivating factor behind their resolution to make a success of their lives, as was the desire to be a good role model and provide a better life for their children. 130,136

What am I supposed to do when my baby has a fever? I had already missed 2 days of school and the teacher told me that if I missed again I would get an F in history. Well, then he got sick, so I got an F.

Teenage mother who received state food aid (USA)129

So I did want to carry on not just to be like a stereotype you know, have kids young, sponging off the dole, sitting at home doing nothing. I wanted to show people that not all young people are just not worth it, they do want to help themselves.

Teenager of Caribbean ethnicity with two (or more) pregnancies (London)130

When I got pregnant, all my family was like you messed up – because I was the smartest one. And now I’m going to college . . . when I’m going to take a step I’m not only taking it for me. I’m taking that step for my daughter.

Teenage mother participating in a school-based intervention (USA)136

Emotional factors: replacing a loss; looking for love

In both studies from the UK,130,134 replacing a loss was a powerful theme. Young women whose first pregnancy had not reached term, as a result of miscarriage or abortion, or whose baby had been stillborn or taken into care, felt an overwhelming sense of loss and bereavement. This was particularly acute when the first baby had been wanted. 130,134 This was described as ‘unnoticed loss’ because the feelings of grief and loss were often not recognised by family, friends or service providers, and therefore these young women received no support or counselling. 130 Under these circumstances, young women intentionally became pregnant again in order to fill the emotional void. Some young women who had lacked emotional warmth in childhood tried to build themselves a family in order to achieve a sense of purpose and fulfilment in their lives, and to feel needed, loved and valued, which had not been provided by their own parents. 130

. . . I never got a chance with him for people to see how capable I was . . . I was just devastated carrying a baby for nine months and feeling it move, going through labour and everything and seeing him for the first time and him just going. It was horrible.

Teenager of Caribbean ethnicity with two (or more) pregnancies (London)130

I was just gonna go straight to the abortion place and my mother said no . . . my dad said . . . you not getting an abortion . . . if I get an abortion I have to leave . . . but in the back of my head . . . I didn’t want an abortion neither.

Teenage mother recruited from social service agency (USA)132

I didn’t go to school a lot because my mum was an alcoholic . . . so instead of me being her baby, she was my baby . . . I suppose I just fell pregnant because I didn’t have no one there with me. I thought I wanted someone to love and take care of.

Teenager of Caribbean ethnicity with two (or more) pregnancies (London)130

If young women’s own abortion decisions were supported, they were more likely to recover emotionally and move on with their lives. If, however, they were encouraged or coerced into having an abortion when they would have preferred to have the baby, they could be deeply distressed and were more likely to go on to have a repeat pregnancy. They would often conceal this pregnancy in order to avoid a second abortion and in order to replace their loss. 130 Furthermore, if young women had been persuaded against their wishes to have an abortion, the feelings of loss were compounded by feelings of guilt. 130,134 It appears that for young women who considered abortion, the choice was not an easy one to make and it was often made with little or no support from family, friends or service providers. If women’s choices were adequately supported, whatever their decision, those who chose to have an abortion would have a better chance of getting their lives back on track, returning to education and avoiding another pregnancy. UK service providers were very concerned about the emotional damage caused by persuading young women to have an abortion against their wishes and/or not providing the appropriate support and counselling before, at the time of, and after the abortion. 130,134 In the absence of adequate support, the emotional pain could sometimes only be alleviated by having another pregnancy and keeping the baby. 130

I felt really bad about what happened last time, when I found out that I was pregnant again I thought I am going to keep it a secret, because I knew that everyone was going to pressure me to have an abortion again, and I didn’t want to do it . . .

Teenager of Caribbean ethnicity with two (or more) pregnancies (London)130

Rationales for rapid repeat pregnancy: creating a family and reverse life-course rationalisation

For some young women, once they had embarked on becoming a mother, they felt it was appropriate to complete their family while still young and energetic, and before moving on with their lives. They felt an only child would be a lonely child and so wanted to provide him or her with siblings. 129 Some second babies were intentionally conceived in order to please a partner and, for some men, fatherhood was proof of their manhood and virility. Young women took this into account in their pregnancy decision-making, reasoning that pride in their achievement might encourage their partner’s commitment to the relationship. 85,129,130 Young women reasoned that they could go back to education and/or a career later on; it seemed to them a more sensible course of action than the traditional choice that other women made to focus on education and work objectives first and then have to interrupt their progress to start a family later, potentially just as their career was taking off. 85,129,130 According to this reverse life-course rationalisation, having one child as a teenager and then delaying completing a family until much later represented the worst of both worlds.

My baby needs a brother or sister – it is too sad to see him growing up without someone to play with.

Teenage mother who received state food aid (USA)129

Now that I’ve had one, I should just finish it, you know, before going back to school and dropping out all over again.

Teenage mother who received state food aid (USA)129

With my baby’s father he was happy when I got pregnant . . . He said it [having a baby] was the best present he could get from me.

Teenage mother who received state food aid (USA)129

I knew what I was doing, I thought he might stick with me if I had a baby so I didn’t want a baby, but I didn’t stop it neither . . . he has kids with other women, so I thought he could with me too . . . if it happened, it happened . . .’

Teenage mother recruited from social service agency (USA)132

Whether the explanation for repeat pregnancies was contextual, motivational, emotional or based on teenage mothers’ own rationale, as described in our findings, it is important to seek an understanding of the complex and diverse reasons that result in some teenage mothers being at a greater risk of repeat pregnancies than others, so that these reasons can be addressed through targeted interventions and the improved provision of services.

Assessment of the certainty of findings

We judged the qualitative studies to be of moderate to high quality; all supported their findings by quoting extracts from the data and in several cases there was evidence of in-depth engagement with participants resulting in particularly rich data sets. We considered whether or not findings were seen in more than one study and in more than one country and, in view of the limited geographical spread of the evidence (seven studies from the USA, two from the UK and one from Australia), whether or not it seemed plausible that they would be transferable between these contexts and to other comparable settings. Overall, our confidence in the certainty of findings was high (for 18 findings) to moderate (for four findings), with only three findings achieving low certainty because they were found in only one study and either the data supporting the finding was relatively thin or the finding itself was equivocal. A table summarising the qualitative findings and indicating the level of certainty for each finding, and a brief explanation of the assessment, can be found in Appendix 17.

Studies identified

We identified 14 RCTs that met our inclusion criteria31,78–93 and two studies that contained useful effectiveness evidence when EPOC criteria52 were applied. One of the randomised trials93 was a cluster randomised trial of poor quality and another was a large study that combined 12 smaller randomised and quasi-randomised studies on one multifactorial intervention into an RCT-style analysis. 78 Both of these studies were judged to have too large a risk of bias to include in the main analysis, but have been included in sensitivity analyses of outcomes as quasi-experimental studies.

Primary analysis

The trial we included in the main analysis examined interventions that fell into two broad types: most were multielement psychosocial programmes and one was a contraceptive programme. The psychosocial programmes offered an array of services, such as case management and referral; education about pregnancy, labour and delivery, contraception, and infant health; child developmental training; contact facilitation with the health-care system; and individual counselling. These programmes were community based, or involved home visits or telephone counselling: seven studies involved home visits,79–82,85,89,91 two studies were community based88,92 and one study used telephone counselling. 90 The contraceptive programme87 offered advanced provision of emergency contraception. The details of these programmes, their lengths and their findings are described in detail in Appendix 19.

The interventions based on home visits had counsellors,138 mentors,90 midwives,78 nurses79 or trained home visitors141 delivering the interventions to young mothers at their homes. These professionals and paraprofessionals could be state sponsored,78 recruited from the community141 or from the same ethnicity. 29,138

It was difficult to establish a definition of usual care or standard practice. Eleven31,80–82,85,87–92 of the 12 studies were from the USA (with the exception being an Australian study)79 where the provision of care for young mothers varies within and between each state. Usual care within each study is described in the study characteristics table.

Ten trials reported the effectiveness of interventions in terms of reducing repeat pregnancy and six trials reported on acceptability (see Appendix 19). The outcomes reported by these 12 studies are outlined in Table 3.

Is there any evidence that supports effective interventions to reduce unintended repeat teenage pregnancies?

Nine of our included studies, involving 1126 young women, 628 of whom were receiving psychosocial interventions and 498 as control, reported the rate of unintended teenage pregnancy, as did the contraceptive intervention study (91 young women participated in this study, of whom 48 received the intervention). The data from psychosocial programmes were split into three subgroups dependent on the delivery of the intervention: home-based (597 participants), community-based (297 participants) and telephone-based (232 participants) interventions.

In none of the home visit studies did the proportion of young women and girls who experienced an unintended repeat teenage pregnancy differ between intervention and control arms. Figure 11 illustrates that, even when these results were pooled, no significant differences could be found between the psychosocial intervention arms and the control arms of the studies and that did not differ by mode of delivery.

FIGURE 11.

Forest plot of comparison: psychosocial intervention vs. usual experience following delivery. Outcome: the proportion of repeat pregnancies (i.e. the effectiveness of the intervention). M–H, Mantel–Haenszel.

The contraception intervention study87 showed a reduction in the number of repeat pregnancies in the intervention group (10 of 48) compared with the control group (14 of 43), giving a RR of 0.69 (95% CI 0.34 to 1.14); however, this was not statistically significant.

Despite these negative findings, four studies31,81,82,85 went on to report the secondary outcome childbirth (75 events from 461 participants). The outcome ‘pregnancies’ covers all cases of known conception. These could have led to miscarriages, abortions, or still or live births. Three of the four studies reporting the outcome childbirth gave RRs of < 1; however, none was significant. Figure 12 illustrates these data.

FIGURE 12.

Forest plot of comparison: psychosocial intervention vs. usual experience following delivery, for the secondary outcome childbirth. M–H, Mantel–Haenszel.

When these results were combined in our meta-analysis, we found a RR of 0.60 (95% CI 0.39 to 0.93) indicating a significant reduction of the number of births observed in the intervention arms of the studies.

Cost-effectiveness of a psychosocial intervention

Only one study investigated the cost-effectiveness of an intervention. 92 This study was a three-arm RCT investigating computer-assisted motivational interviewing (CAMI), with (CAMI+) or without a multipart home visit programme. In this study, the costs of both CAMI+ and any CAMI were evaluated and compared with the control arm.

The perspective of the analysis was not reported in the paper. Only direct intervention costs were included in the analysis, although intervention costs did include start up and implementation determined by programme personnel admin records adjusted to 2009 US dollars. Weighted mean costs were US$2064 per teenager for any CAMI (n = 167), US$2735 per teenager for CAMI+ (n = 80) and US$1449 per teenager for CAMI only (n = 87).

The costs associated with CAMI+ were higher than those for CAMI only because each CAMI+ interventionist carried a smaller caseload than CAMI-only interventionists (20 vs. 60 teenagers) to provide a greater frequency of the enhanced home-visiting components.

Incremental cost-effectiveness ratios (ICERs) were normalised to a sample of 100 teenagers for any CAMI versus usual care, CAMI+ versus usual care and CAMI only versus usual care. The cost per prevented repeated birth was calculated by dividing the cost of the intervention by the number of repeated births between the two groups (intervention and control groups). Unadjusted and adjusted ICERs were calculated. ICERs were adjusted for age, prior birth, continuous insurance, history of sexually transmitted infection and contraceptive-use intentions during pregnancy using logistic regression analyses. Table 4 shows the unadjusted and adjusted ICERs of cost per prevented repeated birth, as calculated by Barnet et al.,83 depending on the level of CAMI received.

TABLE 4 - The unadjusted and adjusted ICERs of cost per prevented repeated birth calculated by Barnet et al.83 depending on the level of CAMI received

Intervention	Unadjusted ICERs		Adjusted ICERs with 95% CI
	Number of repeat births prevented	Estimated cost (US$)	Number of repeat births prevented	Estimated cost (US$)	Lower CI bound (US$)	Upper CI bound (US$)
CAMI only	8	18,672	10	15,078	11,546	21,092
Any CAMI	9	21,895	12	17,388	11,821	31,901
CAMI+	11	24,313	14	19,247	15,085	26,072

Further ICERs were calculated based on a set of scenarios created for specific subpopulations of teenagers with respect to intervention group, age, insurance status and parity. The costs per prevented repeated birth were highest for the youngest teenage mothers in the CAMI+ group (12 years old) with continuous insurance (US$27,187) and lowest for the oldest teenage mothers in the CAMI-only (18 years old) group who were newly insured (US$6822).

This cost-effectiveness analysis was appraised using the Drummond checklist (see Appendix 8). 56 The analysis met certain checklist criteria. The analysis had a clearly defined question and the effectiveness of the intervention was assessed using a RCT. The costs and consequences were justified, credible and appropriate for the research question posed. ICERs were calculated for the two intervention arms and usual care. To test uncertainty further, analyses were conducted based on a set of scenarios created for specific subpopulations of teenagers with respect to intervention group, age, insurance status and parity. The authors considered issues of wider concern by comparing the large costs of unwanted teenage pregnancy on the state, with the costs of prevention, stating that the programme demonstrates good value for money in Maryland, USA. The authors also compared their results with other economic evaluations of teenage pregnancy prevention interventions, although the evidence base is constrained. However, the analysis failed to meet certain checklist criteria: no description of usual care was given; the perspective of the analysis was not stated; the authors considered the direct costs of the intervention, including start up and implementation, yet other costs, such as those to the individual and society, were not considered; no discount rate was reported, although the time horizon of the analysis exceeded 1 year; and the authors stated that the unadjusted and adjusted ICERs presented were normalised to a sample of 100, although no details of the methods and rationale for this normalisation were given. In their conclusion, the authors stated that the results have limited generalisability because of the setting, the population, the geographical area and the short follow-up period of 2 years.

Although the search identified only one economic evaluation, Key et al. 141 stated economic findings in their write-up of the effectiveness of a secondary teenage pregnancy prevention intervention that included school-based social work services co-ordinated with comprehensive health care for teenage mothers and their children. Key et al. 141 proposed that this secondary teenage pregnancy prevention intervention under investigation would result in societal cost savings to the state of approximately US$19,000 for each birth that was postponed until adulthood, which equated to cost savings of approximately US$60,000 for each year of intervention implementation.

Acceptability of interventions

The acceptability of interventions was not directly measured by any study; however, we inferred that if there were any differential dropout between the control and the intervention arms of a study, we might have a suitable proxy for acceptability. We observed that no study, except that of Ford et al. ,88 showed significant differences between arms, and, on combining that with the other trial reporting psychosocial interventions delivered in a community setting (Stevens-Simon et al. ),92 there were no overall significant differences between arms. Figure 13 illustrates these results in a forest plot.

FIGURE 13.

Forest plot of comparison: psychosocial intervention vs. usual experience following delivery. Outcome: the acceptability of the intervention. M–H, Mantel–Haenszel.

Contraception and unprotected sex

Two studies, one reporting a contraception intervention87 and one reporting a psychosocial intervention using trained home visitors,85 reported outcomes related to contraception use and unprotected sex. As these interventions were very different in nature, we did not combine the data in any way. Table 5 shows that the contraception intervention was unsuccessful in increasing either condom use or oral contraception, or in reducing unprotected sex. Although the results are non-significant, it should be noted that all three of these outcomes favour the control arm.

TABLE 5 - Summary of the effects of a contraception programme on other secondary outcomes (Belzer et al.)87

Outcome measure	Number of events		RR	95% CI
	Intervention arm (n = 48)	Control arm (n = 43)	Random effects	Lower bound	Upper bound
Number of repeat pregnancies	10	13	0.69	0.34	1.14
Condom use	15	18	0.75	0.43	1.29
Oral contraception	5	8	0.56	0.56	1.58
Unprotected sex	28	22	1.14	0.78	1.66

In contrast to Belzer et al. ,87 Barnet et al. 85 did show some evidence of increased use of both condoms and oral contraception. While the use of condoms was not statistically greater in the intervention group, use of oral contraception was. Table 6 shows these results. The final comparison in this table is for ‘birth control’ and shows no difference between the study arms.

TABLE 6 - Summary of a psychosocial programme on contraception (Barnet et al.)85

Outcome measure	Number of events		RR	95% CI
	Intervention arm (n = 31)	Control arm (n = 32)	Random effects	Lower bound	Upper bound
Condom use	17	12	1.46	0.84	2.53
Oral contraception	22	14	1.62	1.03	2.55
Birth control	18	21	0.88	0.6	1.31

School dropout

As school dropout is considered to be related to first conceptions, several authors addressed this outcome in their intervention studies. Three studies reported the rates of either successful graduation or continuation in school, or actual dropout. 80,89,95 After re-scaling these variables to read in the same direction, we combined them in a meta-analysis. Barnet et al. 85 demonstrated a RR of 0.52 (95% CI 0.28 to 0.97) in favour of a psychosocial intervention implemented by home visitors to reduce school dropout. Neither of the other two studies replicate this finding; however, Havens et al. 89 did demonstrate marginally fewer dropouts in the intervention arm. When combined, these results demonstrate a RR of 0.74 (95% CI 0.51 to 1.07), which is marginally below the threshold for statistical significance. Figure 14 illustrates the full results of this analysis.

FIGURE 14.

Forest plot of comparison: psychosocial intervention vs. usual experience following delivery. Outcome: school dropout. M–H, Mantel–Haenszel.

Overall completeness and applicability of the evidence

Quality of the evidence

The studies had some overall strengths: most had sample sizes of more than 100 participants, most carried out 24-month follow-ups, most described data collection instruments, most used measures to ensure the validity of self-reported data, most controlled for baseline differences in statistical analyses, and most reported the causes and possible impacts of loss of follow-up.

The risk-of-bias table and figure (Table 7 and Figure 15) highlight that, for many of the studies, it was difficult to judge the methodological quality because of insufficient reporting of key methodological features.

TABLE 7 - Risk-of-bias summary ordered by overall risk of bias: review authors’ judgements about each risk of bias item for each included study

A, random sequence generation, selection bias; B, allocation concealment, selection bias; C, blinding (participants and personnel), performance bias; D, blinding of outcome assessment, detection bias; E, incomplete outcome data, attrition bias; F, selective reporting, reporting bias; G, other bias.

Study	Bias
	A	B	C	D	E	F	G
Koniak-Griffin et al., 200381	+	?	–	+	?	?	+
Barnet et al., 200785	?	–	–	+	+	+	?
Quinlivan et al., 200379	+	+	–	?	?	?	?
Belzer et al., 200587	+	+	?	?	+	?	–
Katz et al., 201190	+	+	–	–	?	?	?
Havens et al., 199789	+	?	–	?	?	?	?
Black et al., 200631	–	–	–	+	+	?	+
Cherniss and Herzog, 199680	?	?	–	+	–	+	–
Sims and Luster, 200291	?	+	–	–	?	?	?
Barnet et al., 200982	+	?	–	–	–	?	?
Ford et al., 200288	?	?	–	–	?	?	–
Stevens-Simon et al., 199792	–	–	–	–	?	?	?

FIGURE 15.

Risk-of-bias graph: review authors’ judgements about each risk of bias item presented as percentages across all included studies.

Four studies80,88,91,95 had insufficient information about the randomisation sequence to make a judgement and just said ‘participants were randomised’ or ‘randomly assigned’; one31 used a randomisation procedure stratified on maternal age and sex of the child; and Stevens-Simon et al. 92 assigned patients to the intervention group using a systematic, non-random approach. This study92 had a high risk of selection, detection and performance bias, and provided insufficient information for attrition and reporting concealed allocation, blinded outcome assessors or separated programme facilitators between the intervention and the control groups. Only one study79 reported adequate allocation concealment using sealed opaque envelopes.

Although these studies reported blinding of study participants and key personnel, it was not emphasised whether this blinding was broken or not. Pragmatic trials cannot use blinding because once randomised, it is impossible to blind the adolescent to her allocated group. Thus, the young woman and the personnel delivering the intervention cannot be blind to the intervention received. However, detection bias was circumvented by some studies that used research assistants to collect outcome data. 88,90,92

Qualitative findings

As there are no linked qualitative studies or process evaluations, we cannot say anything about implementation of included interventions in the review. One of the qualitative studies focused on a specific intervention (the school-based ‘pregnancy free club’),136 but it was not possible to assess the effect of this intervention as no quantitative evaluation has been published. All of the studies contributed data that helped to explain why young women failed to prevent second or subsequent pregnancies and, to some extent, what might have helped them to do so. Themes identified included barriers to understanding about both the nature of fertility and the properties and relative merits of different contraceptive methods; difficulties relating to access to and continuity of contraceptive care; and facilitators of contraceptive uptake, including choice, acceptability, support and encouragement.

Barriers to understanding: knowledge, information and misinformation

Some young women lacked even the most basic knowledge about contraceptive methods. In some cases, they found it difficult or embarrassing to talk to their mothers about sex and contraception even when the mothers were open and willing to discuss such matters;129,139 more open discussion between mothers and daughters sometimes became easier after the first pregnancy, in which case family could be a useful source of information and help. 139 In one US study, young mothers who were enrolled in a combined public health and educational intervention valued the daily presence of a nurse in school because she was considered a useful and easily accessible source of unbiased and authoritative knowledge and information. 136

After delivery, sex appeared to be a more open topic among some of the teenagers’ families . . .  However, some parental communication about sex remained limited . . .

Wilson et al., 2011 (USA)139

I haven’t really talked to my mom [about contraceptives] . . . She just tells me to take care.

17-year-old Latina girl engaged in education/training (USA)139

Focus group participants appeared to be very comfortable talking [with nurses] about birth control methods and issues.

Schaffer et al., 2008 (USA)136

I think also birth control wise sometimes you have questions about – like am I bleeding too much, is this ok? . . . They help you find other resources.

Teenage mother participating in a school-based intervention (USA)136

There were some common misconceptions, particularly about fertility. Some young mothers thought it was not possible to get pregnant soon after a birth, miscarriage or abortion so, consequently, they did not perceive the need for contraception at that time. 130 One young woman, who had been using oral contraceptives when she got pregnant, supposed herself to be particularly fertile and, believing that no contraceptive method could provide adequate protection in her case, declined to use any; she believed that abstinence was the only solution, but was not successful in achieving it and so became pregnant again. 138 Others could not imagine that they would be interested in having sex so soon after the birth130,139 or overestimated the contraceptive effect of breastfeeding. 85 Misconceptions about the side effects of some types of contraception deterred young women from using them. Some of these were spread by rumour (‘the coil can rot your insides’, etc.), but some young women misinterpreted the product information on the packaging of contraceptives or the information given to them by health-care professionals, or focused on rare adverse effects, particularly infertility, and became quite fearful of contraceptive products. 130,131,138 Such misconceptions could lead them to use less reliable contraceptive methods, if any at all. One teenage mother said:

I mean, I’m gonna get pregnant if I have sex. That’s what I tell myself, so . . . I’m just gonna stop altogether. I have to. It’s gonna be hard, but, yeah. I got pregnant with my son on the pill . . .

Teenage mother who expressed an intention to obtain an intrauterine device (IUD) (USA)138

This particular teenager did go on to have another pregnancy. 138 Another teenage mother said:

I used the pill but I stopped, the injections make you bleed, the coil can rot you out and make you stink down below.

Teenager of Caribbean ethnicity with two (or more) pregnancies (London)130

This teenager ‘was so adamant about her views that no amount of attempt to convince her otherwise would have changed her mind’. 130

Additional concerns included having a foreign object inside of one’s body (‘something inside [me] that doesn’t belong for 5 years’) and the string check requirement (‘I don’t think I could do that. Like go up there, like every . . . [period].’).

Weston et al., 2012 (USA);138 study on teenage mothers who expressed an intention to obtain an intrauterine contraceptive device (IUCD)

The most common misconception about abortion was that, as a consequence, a woman would never be able to have a child. This misconception sometimes led directly to a rapid repeat pregnancy. Some young women believed it to be true and so saw no need for contraception. Others, because their anxiety about possible infertility was so acute, got pregnant in order to see whether or not they could, and to alleviate their distress. Information given by health professionals about the minor risks of infection leading to adverse effects on fertility after an abortion appeared to be overemphasised. This could have been with the intention of discouraging young women from putting themselves into the position of needing an(other) abortion, or, possibly, it was misinterpreted as a major threat. These fears were sometimes reinforced in schools where, in both sex and relationship education (SRE) and religious education, abortion was presented from a negative, moralistic stance. 130,134

After the second abortion my doctors told me that my chances of having kids would be slimmer . . . I was being a bit more careful thinking that I won’t get pregnant again, but at the same time sort of hoping that I would be, because then I would know that I could have children.

Teenager of Caribbean ethnicity with two (or more) pregnancies (London)130

. . . if doctors and nurses and other people spent less time telling women they were risking their fertility by having an abortion there would be less second abortions . . . I see people back pregnant just a few months after an abortion and they are really surprised.

Hoggart et al., 2010 (UK)134

One problem we do find, is that some of the girls have had some quite heavy handed sex education . . . they’ve obviously been exposed to something fairly dramatic in schools which has had quite an impact in terms of putting them off abortion.

Hoggart et al., 2010 (UK)134

Barriers to contraceptive uptake: access and continuity of care

For some young women, contraceptive counselling given during pregnancy, or immediately afterwards, together with prescription, facilitated timely uptake of effective contraception. These young women tended to have more frequent contact with the health-care system and received more information and counselling about contraceptive methods, particularly during pregnancy and shortly afterwards. They were often supplied with their method of choice at the hospital after delivery. 139 In contrast, other young women encountered significant barriers related to difficulties in accessing contraceptive care or the contraceptive method of choice, including missing appointments because of restrictive clinic hours, the requirement to obtain a referral letter, the requirement to be aged 18 years or over or have parental consent, health requirements [e.g. to remain sexually transmitted disease (STD) free for the preceding year], lack of provider training to supply the contraceptive device of choice (e.g. to fit an IUCD or hormonal implant) and lengthy gaps between requesting a device and having it fitted. 138,139 In the USA, lack of Medicaid cover was also a problem. 139 In the Australian study,137 however, it was stated that all the participants, whether they were pregnant, were parenting or had experienced a recent abortion, and whether the first pregnancy had been planned and wanted or unintended and unwanted, expressed firm intentions to be vigilant about their contraceptive use in the immediate future. Those who had had abortions experienced a heightened sense of pregnancy risk and consequently were keen to avoid repeating the experience; those who continued the pregnancy wanted to delay further childbearing.

They were just saying about the pill, the shot, like just different methods. And I just asked, ‘Could I get the pill?’ And they just prescribed me some pills.

Teenage mother engaged in education/training (USA)139

Barriers included limited clinic hours . . . ‘I missed my appointment . . . last month but I couldn’t do it ‘cause I couldn’t afford to miss any more days out of school.’

Weston et al., 2012 (USA);138 study on teenage mothers who expressed an intention to obtain an IUD

I’ve basically made the decision, like, for when the baby’s born, I’m getting straight back on the pill but I’m getting a different one and he’ll be wearing a condom as well ‘cause I’m not taking any more chances after this, I’ll be really cautious.

Smith et al., 2013 (Australia);131 study on pregnant, parenting and post-abortion teens

However, key informants in the UK were concerned about a lack of post-abortion follow-up care for teenagers that included establishing an effective contraceptive regime and addressing any social problems that had contributed to the girl getting pregnant. This could be associated with commissioning issues: primary care trusts varied regarding whether or not they commissioned contraceptive services at the time of abortion and whether or not this included LARCs. Continuity (including having referral, assessment, procedure and follow-up components all in place) was thought to be important so that intervention opportunities would not be missed. It was considered most important to use the referral consultation to begin supporting contraceptive choices after the abortion. Where continuous services were not available, young women could be left at risk of a repeat pregnancy. 134 According to Hoggart et al. (2010),134 one of their study participants, Yasmeen, ‘had plans for the future, and she wanted to ensure that she did not become pregnant again. She would have liked to have an injection at the clinic . . .’

They are sending you somewhere else, and I don’t even know where that is.

Teenager who had had an abortion (UK)134

Other findings of interest

Conceptualising the issue

There was a polarisation in the evidence base with regards to the epistemology of interventions to prevent or delay subsequent teenage pregnancy. Authors of some of the quantitative evidence viewed teenage pregnancy from the perspective of irresponsible, problem behaviour, typically associated with alcohol use, socioeconomic deprivation and particular ethnic groups (in the USA) which are considered at high risk. 31,98–101,157 Conversely, writers of the qualitative reports viewed teenage pregnancy from the perspective of the adolescents themselves or health-care professionals who work closely with young mothers. From this perspective, sexual activity was often thought of as spontaneous, unplanned and sometimes involuntary. 85

Interventions developed from this issue, being based on the informed decision-making of young women who weigh up the consequences of pregnancy and motherhood, may be unrealistic. 85,154 Stakeholders agreed that for some adolescents there was no planning or consideration of consequences, and no planned behaviour: adolescents would prioritise carrying their mobile phone, but not contraception. Generally, it was felt that adolescents live for the moment and do not believe that they will get pregnant.

Evidence suggests that viewing repeat teenage pregnancy from a developmental approach, focusing on the individual’s level and abilities, cultural context, age-appropriate impulsive and rational decision-making styles, may be a better place to start with regard to conceptualising interventions. 85

The NHS Scotland strategy document Reducing Teenage Pregnancy142 states that promoting a culture of participation is one of many effective methods for reducing teenage pregnancy. In line with UNICEF’s UN Convention on the Rights of the Child,158 more successful interventions might be those that involve young people at a level that is both meaningful and appropriate. This culture of participation includes creating opportunities for young people’s individual experiences of education and services related to sex and relationships to be heard and influence their future direction. Opportunities to involve young people should be considered throughout the development and delivery of work, for example in scoping objectives, creating leaflets and marketing campaigns. 154,159 Framing the issue using adolescent views and opinions, rather than those of adults, and by focusing on their control and abilities to manage sexual relationships through engaging in discussion might provide conditions under which interventions will work because there is the opportunity for the young people to take responsibility for actions and choices.

Community

A community may value rapid repeat teenage pregnancies, believing motherhood to be a desirable and defining role for an adolescent. 31 Communities also play an important role in how pregnancies are perceived; stakeholders stated that in more deprived areas, pregnancy is viewed as normal and almost expected, while in less deprived, more privileged areas, teenage pregnancy is viewed as out of the ordinary and, in certain circumstances, shameful. The differing perceptions of teenage pregnancy could lead to differing ages of onset of sexual activity, differing use of contraception and different decisions regarding the outcome of pregnancy, such as termination, adoption or raising the child, depending on what the community regards as ‘normal’ or ‘acceptable’. 31 Aligned to the expectations of a person’s community, the teenager’s culture provides a context for sexual activity, use of contraception and terminations. This context might influence an adolescent’s decision-making and their opinions on the potential consequences of pregnancy, affecting their behaviour and, in some cases, limiting their options if they were to become pregnant unintentionally. 99–101,156,157

Communities can also hinder or facilitate intervention delivery. Stakeholders working closely with young mothers cited an example of an area in which there is little or no engagement in available services for young mothers. The area in question is one with a strong sense of community. Therefore, it might be more challenging to operationalise interventions in close, tight-knit communities in which service providers are viewed as ‘outsiders’; in such circumstances, attention should be paid to developing approaches and strategies from the ground up.

Family

The importance of understanding a teenager’s family context is also important. Stakeholders suggested that a difficult home life caused by, for example, disagreements with family members or a breakdown in family structure, as often experienced by looked-after children, can lead to the desire to create a family of one’s own, as a way of dealing with the feelings of family loss experienced by these adolescents. Herrman133 also suggested that pregnancy and the birth of a child may meet the needs of the adolescent, as they perceive this to enhance family closeness.

Teenage childbearing can also run in families, thereby influencing future generations. 98 If their own mothers had been teenage mums, teenage girls are more likely to experience teenage pregnancy, perhaps because childbearing is normalised from an early age and a large emphasis is placed on motherhood. If teenagers have helped to raise younger siblings, they may feel competent enough to raise children of their own, thus altering their perception and reducing the gravity of the consequences of becoming pregnant themselves. A mother attending a service user feedback group conducted by the study team stated that she was one of 10 children and had helped to raise her siblings from a very young age. When she had her first baby, she said it was no different: ‘I had basically been a mother since the age of 10 anyway.’

These findings point to the need to better understand the motivations of a teenager with regard to becoming pregnant, as these might be highly complex, and emotionally and socially embedded and therefore not necessarily visible to those providing services and support.

Peer influence

Peer influence can provide a context in which a teenager feels they do not fit if they are not sexually active or have not experienced a pregnancy. 97,129 Peer influence can be a significant driving force for sexual activity, as all teenagers want to fit in with their peer group.

Therefore, interventions that operate within the context of teenagers’ peer groups might have more impact than those that operate in isolation of this context.

Partner influence

The relationship between the young women and their partners also provides a further layer of context, particularly for the risk of rapid repeat teenage pregnancy. Stevens-Simon et al. 155 reported that the following partner variables were associated with repeat pregnancy: having an older boyfriend, having a new boyfriend since the conception of the first pregnancy and having a boyfriend who wanted a child. The main reason highlighted, as regards partner influence, in the evidence was the ability of the partner to influence child bearing intentions and contraception use. 97,99–101,155 Chapter 3 describes how partners and young women’s relationships with them influence contraception intentions and use, and views of pregnancy.

In summary:

• Older partners may exert influence over an adolescent because of the adolescent’s limited experience and confidence; the adolescent trusts that their partner knows better than themselves.

• Gender roles and cultural influences place women in passive roles, unable to state whether or not sex will occur and whether or not contraception will be used. 97,99–101

• Those in long-term relationships with new partners since the birth of their first baby, may become pregnant again to give the partner a baby of their own as a way to repay them for helping to look after the older child, and to cement their relationship (stakeholders from the first meeting). Stakeholders also stated that young mothers want to give their new partner a baby and perceived this as a way of ‘keeping him’.

Female adolescents in relationships with older partners tend to be younger, less likely to use a condom at first intercourse, less likely to use condoms consistently and more likely to become pregnant than adolescents with partners of similar age . . .

Raneri and Wiemann (2007)97

Previous experiences

Relationships and a partner’s influence can have a powerful effect and, therefore, the effectiveness of interventions may be mediated by this context. This issue also highlights the need to pay attention to the context around the teenager when designing and delivering interventions.

Summary

There are complex layers of context that have the potential to influence the motivations, intentions and behaviours of young women, with respect to sexual activity and pregnancy. These different contexts provide both an explanation and some signposting as regards what should be accounted for in the development and implementation of interventions so that they have the desired impact. Our analysis of the evidence leads us to suggest that ‘framing’ the issue from the perspective of the young person, which includes their active engagement in strategy development and implementation, may provide a more conducive context for an intervention to work. Furthermore, paying attention to the potential ways that communities, families, peers and partners can both be engaged in and influence intervention development and use could help situate the intervention in a context that is relevant to the individual. We also draw attention to the emotional aspects of the issues surrounding sexual activity, relationships and pregnancy (particularly in the context of previous losses), which may be more or less visible to those providing services, but which provide a powerful backdrop to the young person’s motivations, intentions and behaviours.

Motivations

As described, different individuals will have different reasons for becoming pregnant; thus, attending to underlying motivations within the development of interventions would be a good starting point. These motivations are driven by community norms, social context, peer influence and past experiences, which all add to a complex mix. Working with these different motivations could lead to better engagment with the issue of sex and pregnancy, and enhance the potential of the young person responding to an intervention and making informed choices.

The motivation(s) to get pregnant – whether these are positive or negative – may be contingent on the issue of how young women plan for or view the future. Evidence suggests that some perceive a dichotomous choice between motherhood31,85,97,130 and other goals and aspirations. 31,91 In the absence of other goals and aspirations, some adolescents look to motherhood. Low educational attainment and disengagement with school may lead teenagers to believe that childbearing may provide them with an opportunity to experience success, gain autonomy and self-esteem that they cannot gain through academic and vocational routes. 31,91 Our stakeholders raised the concern that parenting skills and praise during pregnancy and the early years may be the only time that a teenage mother has felt as though she is good at something. Furthermore, in low-income communities, it has been observed that there may be ‘rewards’ (such as respect in the community) for bearing children, where the norms might not discourage having children at a young age. 85

Adolescents can also have low expectations of life. Our stakeholders suggested that teenagers may get pregnant as there are no jobs and no prospects. Girls sometimes ‘drift’ and are unsure of what to do when they find themselves pregnant, or they could be unaware that they are pregnant. In this context, they continue with pregnancy because they perceive that they have limited options.

There was also evidence to suggest that some adolescents perceive having children during their teenage years as a sensible and beneficial decision, which was coined by Clarke130 as a ‘reversed life course’. By having their children young, mothers have the rest of their lives to focus on a career; once their child is old enough to attend school full time, the majority of mothers who became pregnant as teenagers are still in their teenage years or early twenties. However, if they wait until they are in their mid to late thirties to begin a family, after they have established a career, it might be difficult or impossible to conceive. From this perspective, unintentional pregnancy is not viewed negatively, but as a positive decision to have children at a young age in order to provide more opportunities and choices in the future.

Therefore, interventions that focus on reducing repeat pregnancy should be considered within the broader context of the young person’s life, motivations, attainment and ambitions. It is important to note that the narratives of some may run counter to the policy discourse, in that pregnancy might have an ascribed positive meaning, which may also be contextually (within their community) accepted as the norm. Furthermore, interventions might usefully focus on alternative strategies (if appropriate) for achieving personal automony and acheivement other than pregnancy.

Taking control

Understanding and working with the different motivations that underpin decisions related to pregnancy could facilitate an adolescent’s ability to take (more) control. Designing and delivering interventions that provide opportunities for adolescents to take control of sexual encounters, contraception and decisions about pregnancy might empower them to make informed choices. This mechanism may, however, be mediated by culturally determined attitudes that have shaped adolescents’ identities and perceptions of sexual roles for males and females, which can also impact on their abilities to negotiate the use of contraception with partners. 97,99–101

Paying attention to self-esteem, self-efficacy (confidence in one’s ability to take action) and empowerment were all stated as ways that could increase the likelihood of an adolescent taking control. 31,81,85,105,129 Our stakeholders also highlighted the need for adolescents to increase their self-esteem, life skills and empowerment. They theorised that if adolescents feel that their options are limited and have low expectations of life, they may also believe that their choices do not impact on their life, leading to more impulsive decisions without thinking through the consequences; these views are also supported by research. 85,154 Policy documents from the UK8,142 also recognise the importance of raising self-esteem and empowerment as part of their strategies to reduce teenage pregnancy. Reducing Teenage Pregnancy (NHS Scotland, 2014)142 also highlights the roles that a wider context, such as the media, can have on influencing behaviour and promoting positive role models.

There are a range of theories that conceptualise the multiple levels of influence that an adolescent is under and that could help to build self-esteem, self-efficacy and empowerment. For example, Raneri and Wiemann97 used social–ecological theory as a basis to explain the influence of these multiple levels (e.g. individual, dyad, family, peer/community and social system) on behavioural outcomes. According to this theory, development and behaviour are influenced by a range of factors, from those that are closest to the individual, such as individual experiences and attitudes, to those that are part of the social environment, such as race, ethnicity and socioeconomic status. Based on this analysis, Raneri and Wiemann97 suggested that multifaceted interventions that target adolescent mothers and the communities in which they live, at delivery and during the early postpartum period, can help them to overcome outcomes such as repeat pregnancy. It was suggested that interventions might include policies, programmes and funding to intensify and lengthen services for young mothers and their children in the postpartum period, to strengthen connections with educational and occupational opportunities, to improve adolescents’ and their families’ understanding of the effects of partner selection and characteristics on their life course and to provide individual and partner health education or counselling on family planning and healthy relationships.

Mbambo et al. 106 drew on the Health Belief Model to provide a context for a teenage girl’s personal circumstances and stated their perceived susceptibility to getting pregnant again, the perceived severity of getting pregnant again, such as the likelihood of a breakdown in family relationships, and the perceived benefits of not getting pregnant again, such as the ability to complete school. In this study, 107 adolescent mothers, attending two clinics in Mkhondo, South Africa, were given a questionnaire that enquired about their biographic information and individual perceptions, and modifying factors and variables that could affect the likelihood of utilising contraceptives, such as age and educational attainment. The application of this theory resulted in the analysis described in the following sections, which may provide a useful heuristic guide to intervention development within specific contexts.

Situating the intervention

As previous sections and findings indicate, the issue of adolescent pregnancy involves family, societal, medical and educational components; therefore, interventions need to address the complex needs of the young women,107 that is not just the issue of early (repeat) pregnancy. Therefore, by paying attention to these wider issues, interventions would be situated within a wider context for the young person.

Failure to address the wider social and cultural influences on teenage pregnancy has contributed to the lack of progress made in reducing it in Scotland in the past . . . The most successful programmes therefore are multi-component and address a wide range of determinants of risk.

Reducing Teenage Pregnancy, NHS Scotland, 2014142

Evidence suggests that this could include medical and social service providers helping adolescents to think through the wider repercussions of childbearing: how their age, educational status (current level of schooling and qualifications), financial situation and relationship status may impact, not only on decisions affecting childbearing but also upon other areas of their life. 97 Interventions should take account of the individual’s internal/external social competence (their ability or inability to handle their inner and outer worlds),81 their community31 and their cultural99,100,156,157 and peer group norms. 97,129 Other relevant questions include those related to previous experiences of pregnancy, including how decisions were made about the outcomes (e.g. whether or not decisions were made for the pregnant adolescent), which might lead to beginning a ‘journey of search’. 130

These wider contextual issues have the potential to influence the outcome of an intervention, for example by making adolescents equate contraception use with obtaining the lifestyle they want. 161 However, they could also help to shape the intervention itself by situating it in the context of the young person, that is by providing elements which may be desired by the adolescents, such as skills development, and re-engagement with education or employment training. 99–101,108,129

The role of education in this group of individuals seems to be important; for example, Berenson and Wiemann108 reported that reliable contraception was observed to occur more frequently in adolescents who were doing well at school and who were enrolled in school within 6 months of delivery. Drayton et al. 99–101 also found that participation in their programme, the Women’s Centre of Jamaica Foundation Programme, exerted a protective effect against the occurrence of repeat pregnancy with key positive variables including the desire to continue education. This programme provides classroom instruction and support for adolescent mothers, aged 16 years and under, with the aim of returning them to the formal school system, and delaying the birth of a second child. The programme also provides skills training and job placement assistance to unmarried, unemployed mothers over the age of 18 years. Acceptance and use of a family planning method is a requirement for programme participation. The theory is that completing education is a motivator for not getting pregnant, and links to issues around achieving hope for the future.

The promotion of educational attainment needs to include practical considerations through, for example, a greater awareness of the challenges that young mothers face when trying to continue their education while parenting. 129 Difficulties include arranging childcare if they are the sole parent, inability to meet deadlines because of limited time to complete assignments, not being able to give tasks their undivided attention and issues of absences if their children are unwell.

Young mothers attending a service user feedback group stated that they wished to continue education, but the needs of their children would require a crèche, and for the intervention to be delivered at school. In the UK, this is currently not an option; however, evidence from the USA demonstrates positive results when interventions are delivered in schools. 97,107,109,162 This approach embeds the intervention in the mother’s lives, keeping the focus on education, but also allowing them to have contact with their children during the day. Providing childcare in a school as part of a wider intervention acknowledges the dual role of the adolescent mother; she is still an adolescent who needs to complete her education, but she is a mother who feels responsible for her child and needs to know that they are safe and well cared for while she is attending classes. 162

Situating the intervention in the broader context of the young person also means considering vulnerable and high-risk groups, such as victims of violence. Stakeholders attending a meeting in Shrewsbury stated that sexual exploitation can be a problem, particularly older men targeting vulnerable, looked-after children. They suggested that there should be appropriate sex education and services specifically for vulnerable children (e.g. looked-after children or victims of violence), because ‘basic’ services are not good enough in these cases. Because of an upheaval in family life and traumatic events in their past, such adolescents may drift and feel their options are limited. There is a risk that these young women will be exploited by men who promise them a stable, loving relationship, only to be manipulated as the relationship continues; with a lack of family, these adolescents could easily find themselves trapped in such relationships with no one to turn to. Policy documents have also highlighted the need to target vulnerable groups and undertake needs assessments in order to provide the support that these adolescents need. 8,142,163

The outcome of ‘situating the intervention’ within the broader context of the young person’s life and taking a wider view of pregnancy in adolescence may lead to a greater potential for ‘connectedness’ with the intervention,31,89,162,164–166 the issue108 and their own context90 and, therefore, the notion of ‘support’ may trigger their self-determination and active control. Feeling connected and supported can help an adolescent to feel that their life choices are being encouraged. 31,166 A supportive professional delivering the intervention, the group itself or family members can help to verbalise and confirm a mother’s skills,108 and help them to develop strategies and plans to change their attitudes and behaviour. 89 Feeling connected and supported as part of a group of mothers with whom they can relate to may inspire mothers to achieve and not feel limited because of their current situation. 166 Young mothers attending a service user feedback group also stated that being part of a group allowed them to hear others’ feelings and opinions, and let them know that they were ‘not alone’. Havens et al. 89 also found that the bond and the supportive relationship between a mentor and an adolescent could give pregnant teenagers the support that they needed to change attitudes and behaviours.

If an adolescent has limited support from family and friends, the connections made as part of an intervention could fill the gaps in their social networks, offering an avenue of support for adolescent mothers that they do not have in their day-to-day lives, which, without the intervention, would have remained unfulfilled. 162 Feelings of support and connectedness can increase self-esteem, self-competence and empowerment, which, as stated previously, can in turn lead to greater feelings of control over self, choices and outcomes. Being connected to the intervention, group or others can also have the added benefit of helping to keep mothers on track with the changes they propose to make. 166 Professionals delivering the intervention or group leaders are ideally placed to routinely check the adolescents’ progress.

Evidence from stakeholder meetings suggests that other opportunities and services (aside from specific interventions) from local government and authorities, such as additional visits from health-care professionals, access to nursery places and community playgroups, may enhance connection to the system, community and other young mothers. It allows young mothers to know that the system still cares, although there seems to be a need for a balance of providing support through different avenues, without the mothers relying on these services and having subsequent pregnancies in order to still qualify for them. Stakeholders attending a meeting in Shrewsbury gave stories of a service that provided young mothers with support through groups and crèche facilities until the child was 2 years old. Mothers enjoyed the service; they built friendships, gained support, felt as though they could share problems and issues, and gain advice in a nurturing environment. Once the child reached the cut-off age of 2 years, the mothers said that they felt lonely because they had become used to seeing their group leaders and the other mothers. In short, they felt abandoned and without support. Some mothers confessed later to having subsequent children in order to go back to the service, reconnect with their peers and regain the lost support.

Additional databases used for grey literature

OpenGrey.

Scopus.

Scirus.

Social Care Online [via Science Citation Index Expanded website].

National Research Register.

NIHR Clinical Research Network Portfolio.

Index to THESES.

Google.

Databases searched for economics literature

RePEc (http://repec.org/).

EconLit.

Databases searched for qualitative literature

ASSIA.

BNI.

SocAbs.

SSCI.

MEDLINE

1. exp pregnancy in adolescence/

2. Pregnancy, unplanned/

3. Pregnancy unwanted/

4. Abortion, induced/

5. Abortion, legal/

6. Or/2-5

7. Adolescent/

8. (Adolescen$ or teen$ or girl or girls or youth or youths or youthful or young or juvenile).ti,ab.

9. 7 or 8

10. (Repeat or subsequent or second or secondary or further or additional).ti,ab.

11. 1 and 10

12. 6 and 9 and 10

13. 11 or 12

14. Limit 13 to 1995 onwards and humans only

PsycINFO

S1 SU.EXACT(“Adolescent Pregnancy”)

S2 SU.EXACT(“Induced Abortion”)

S3 AB,TI(Adolescen* OR teen* OR girl or girls OR youth or youths or youthful OR young OR juvenile)

S4 AB,TI(Repeat OR subsequent OR second OR secondary OR further OR additional)

S5 S1 AND S4

S6 S2 AND S3 AND S4

S7 S5 OR S6

S8 Limit to 1995 to 2013

Cumulative Index to Nursing and Allied Health Literature

S1 (MM “Pregnancy in adolescence+”) OR (MM “Adolescent Mothers) OR (MM “Adolescent Parents+”)

S2 (MM “Pregnancy, unplanned”) OR (MM “Pregnancy unwanted”) OR (MM “Abortion, induced”)

S3 TI (Adolescen* or teen* or girl or girls or youth or youths or youthful or young or juvenile)

S4 AB (Adolescen* or teen* or girl or girls or youth or youths or youthful or young or juvenile)

S5 S3 OR S4

S6 TI (Repeat or subsequent or second or secondary or further or additional)

S7 AB (Repeat or subsequent or second or secondary or further or additional)

S8 S6 OR S7

S9 S1 AND S8

S10 S2 AND S5 AND S8

S11 >S9 OR S10

S12 Limit 11 to yr=”1995-current”

Cochrane Library

1. Pregnancy in adolescence/

2. Pregnancy, unplanned/

3. Pregnancy, unwanted/

4. Abortion, legal/

5. Abortion, induced/

6. OR/ 1-5

7. Adolescent/

8. (adolescen* or teen* or girls or girls or youth or youths or youthful or young or juvenile):ti,ab.

9. 7 OR 8

10. (repeat or subsequent or second or secondary or further or additional):ti,ab.

11. 1 AND 10

12. 6 AND 9 AND 10

13. 11 OR 12

14. Limit 13 to 1995 onwards

EMBASE

1. exp adolescent pregnancy/

2. unplanned pregnancy/ or unwanted pregnancy/

3. legal abortion/ or induced abortion/

4. 2 or 3

5. adolescent/

6. (Adolescen$ or teen$ or girl or girls or youth or youths or youthful or young or juvenile).ti,ab.

7. 5 or 6

8. (Repeat or subsequent or second or secondary or further or additional).ti,ab.

9. 1 and 8

10. 4 and 8 and 9

11. 9 or 10

12. limit 12 to 1995 onwards and humans only

British Nursing Index (BNI)

S1 SU.EXACT(pregnancy : age factors)

S2 SU.EXACT(“Abortion”)

S3 SU.EXACT(“Adolescents”)

S4 AB,TI(Adolescen* OR teen* OR girl or girls OR youth or youths or youthful OR young OR juvenile)

S5 S3 OR S4

S6 AB,TI(Repeat OR subsequent OR second OR secondary OR further OR additional)

S7 S1 AND S6

S8 S2 AND S5 AND S6

S9 S7 OR S8

S10 Limit S9 to 1995-2013 and humans

Educational Resources Information Center

S1 SU.EXACT(“Early Parenthood”)

S2 AB,TI(Abortion)

S3 SU.EXACT(“Adolescents”)

S4 AB,TI(Adolescen* OR teen* OR girl or girls OR youth or youths or youthful OR young OR juvenile)

S5 S3 OR S4

S6 AB,TI(Repeat OR subsequent OR second OR secondary OR further OR additional)

S7 S1 AND S6

S8 S2 AND S5 AND S6

S9 S7 OR S8

S10 Limit S9 to 1995 onwards and humans

Sociological Abstracts

S1 SU.EXACT(“Adolescent Pregnancy”) OR SU.EXACT(“Adolescent Mothers”)

S2 SU.EXACT(“Abortion”)

S3 SU.EXACT(“Adolescents”)

S4 AB,TI(Adolescen* OR teen* OR girl or girls OR youth or youths or youthful OR young OR juvenile)

S5 S3 OR S4

S6 AB,TI(Repeat OR subsequent OR second OR secondary OR further OR additional)

S7 S1 AND S6

S8 S2 AND S5 AND S6

S9 S7 OR S8

S10 Limit S10 to 1995 onwards and humans

BiblioMap

1. Category: Focus of the report: pregnancy prevention

2. Freetext: pregnancy

3. Freetext: pregnancies

4. Freetext: pregnant

5. Freetext: conception

6. Freetext: conceptions

7. Freetext: abortion

8. Freetext: abortions

9. OR/1-8

10. Category: Characteristics of the study population: young people

11. Freetext: teen

12. Freetext: teens

13. Freetext: teenage

14. Freetext: teenagers

15. Freetext: adolescent

16. Freetext: adolescents

17. Freetext: adolescence

18. Freetext: youth

19. Freetext: young

20. Freetext: youthful

21. Freetext: juvenile

22. Freetext: girl

23. Freetext: girls

24. OR/10-23

25. Freetext: repeat

26. Freetext: subsequent

27. Freetext: second

28. Freetext: secondary

29. Freetext: further

30. Freetext: additional

31. OR/25-30

32. AND 24 AND 31

Applied Social Sciences Index and Abstracts

S1 SU.EXACT(“Pregnant adolescent girls”) OR SU.EXACT(“Adolescent mothers”)

S2 SU.EXACT(“Abortion”)

S3 SU.EXACT(“Adolescents”)

S4 AB,TI(Adolescen* OR teen* OR girl or girls OR youth or youths or youthful OR young OR juvenile)

S5 S3 OR S4

S6 AB,TI(Repeat OR subsequent OR second OR secondary OR further OR additional)

S7 S1 AND S6

S8 S2 AND S5 AND S6

S9 S7 OR S8

S10 Limit S10 to 1995 onwards and humans

Social Sciences Citation Index (Web of Science)

Free text search:

1. Pregnancy

2. Pregnancies

3. “conception*”

4. “abortion*”

5. OR 1-4

6. “teenage*”

7. “adolescent*”

8. 6 or 7

9. Repeat

10. Subsequent

11. Secondary

12. OR 9-11

13. 5 AND 8 AND 12

14. Limit to 1995 onwards

OpenGrey

“teenage pregnancy” OR “teenage pregnancies” OR “teenage conception” OR ” “teenage conceptions"

Scopus

Teenge pregnancy AND repeat

Scirus

“repeat +pregnancy” AND (teenage* OR adolescen*)

Limit to 1995 onwards and excluding journal sources

Social Care Online (via Science Citation Index Expanded website)

Teenage pregnancy AND repeat

Limit to 1995 onwards

National Research Register archive

Keyword search: adolescence and pregnancy

National Institute for Health Research Clinical Research Network Portfolio

Reproductive health (all section screened)

Index to THESES

Title search: teenage pregnancy or teenage pregnancies or teenage abortions or teenage conceptions

Google

Site:uk “repeat pregnancy” teenager

Site:uk “repeat pregnancy” teenagers

Site:uk “repeat pregnancies” teenagers

Site:uk “repeat pregnancies” teenager

Site:uk “repeat pregnancy” teenage

Site:uk “repeat pregnancies” teenage

Site:uk “repeat pregnancy” teen

Site:uk “repeat pregnancies” teen

Site:uk “repeat pregnancy” teens

Site:uk “repeat pregnancies” teens

Site:uk “repeat teen” pregnancy

Site:uk “repeat teen” pregnancies

Site:uk “repeat teenage” pregnancy

Site:uk “repeat teenage” pregnancies

Site:uk “second pregnancy” teen

Site:uk “second pregnancies” teen

Site:uk “second pregnancy” teens

Site:uk “second pregnancies” teens

Site:uk “second pregnancy” teenage

Site:uk “second pregnancies” teenage

Site:uk “second pregnancy” teenager

Site:uk “second pregnancies” teenager

Site:uk “second pregnancy” teenagers

Site:uk “second pregnancies” teenagers

Research Papers in Economics

Free text search:

(teen | adolescent | adolescence | girl | young | youth | juvenile) + (pregnancy | pregnancies | conception | abortion) + (repeat | subsequent | second | secondary | further | additional)

Limit to 1995 onwards

American Economic Association’s electronic bibliography (EconLit)

1. SU = pregnancy and teens or SU = adolescent mothers or SU = adolescent pregnancy

2. SU = pregnancy or SU = abortion or SU = conception

3. AB = (adolescen* or teen* or girl* or young or youth* or juvenile)

4. AB = (repeat or subsequent or second or secondary or further or additional)

5. 1 and 4

6. 2 and 3 and 4

7. 5 or 6

8. Limit 7 to 1995 onwards

Morning session: presentations by the review team to the stakeholders

Lunch and networking

Afternoon session

Participants were divided into three groups for discussion and feedback of three each of the key questions, listed below.

Introduction, background and study update

Rhiannon Whittaker (chief investigator) thanked the Task and Finish Group for facilitating this meeting with stakeholders and thanked everyone for attending. She gave an overview of the objectives of the session and added that comments and questions would be requested at each stage of the meeting. These would be noted and circulated to all later with the report of the meeting and with a request for comments on accuracy or omissions. Following the presentations, the team had some questions to pose to stakeholders.

Overview of study methods

Maggie Hendry (lead reviewer) explained the research brief and outlined the mixed-methods approach to the study with a brief outline of systematic review methodology. Summary of review methods:

• Phase 1

  • literature searches, inclusion/exclusion, quality appraisal and mapping exercise

  • consideration of mapping exercise and consultation with advisory group and stakeholders.

• Phase 2

  • selection of records for in-depth review and data extraction.

• Phase 3

  • synthesis of evidence according to type using design appropriate tools.

• Phase 4

  • overarching narrative synthesis and interpretation.

The results presented at this meeting are the Phase 3 results.

Overview of mapping exercise and final numbers of included studies

Rabeea’h Aslam (INTERUPT project manager and systematic reviewer) gave an overview of the mapping exercise and explained how papers were sifted using CART criteria to achieve the final number of included studies. Details of intervention types identified and outcomes measured were given.

A total of 118 studies were identified for the mapping exercise. Through the process of discussion and development of the CART criteria, the team culled the final number of papers to 12 RCTs, 10 qualitative studies and 49 non-randomised observational studies.

Components for complex interventions:

1. pregnancy testing and maternity counselling

2. adoption counselling and referral services

3. primary and preventative health-care services (including prenatal and postnatal care)

4. nutrition information and counselling

5. referral for screening and treatment of sexually transmitted infections, including HIV/AIDS

6. referral to appropriate paediatric care

7. educational services related to family life and problems associated with adolescent premarital sexual relations

8. appropriate educational and vocational services

9. mental health services

10. counselling and referral for family planning services

11. job skills training

12. tutoring

13. mentoring

14. addressing domestic violence and peer-relationships

15. supportive counselling and social work services.

Outcomes measured

Primary outcomes:

• identification of at-risk group

• identification of barriers to and facilitators of interventions relating to acceptability, uptake and feasibility of implementation

• views and experiences of young women, families and professionals.

Other outcomes:

• identification of, access to and uptake of reproductive services and social care

• feasibility of widespread adoption of interventions in health and social care

• cost of interventions

• effectiveness of interventions

  • change in validated quality of life indices

  • change in rate of abortion

• cost-effectiveness of interventions.

Overview of quantitative data synthesis

Ben Carter (statistician) gave an overview.

The interventions considered were:

• control group (usual care or usual experience)

• counselling (home visit, community, telephone or in person)

• contraception.

The primary outcomes considered were:

• acceptability of interventions

  • proportion of participants who reported that the intervention was acceptable or, in the absence of this, the proportion of participants who were willing to be recruited into the study.

• effectiveness of interventions

  • proportion of unintended repeat teenage pregnancies.

• uptake of the interventions

  • proportion of participants who were recruited and allocated to the intervention, compared with those recruited.

Summary of findings

Phase 3: qualitative results

Maggie Hendry (lead reviewer) presented the qualitative results from the study.

We included nine studies, published between 1998 and 2012, seven of which were conducted in the USA, one in the UK and one in the UK and the Caribbean. Participants in the US studies were predominantly African American or Hispanic, UK participants were of mixed ethnicity, but all were in London. Only one study addressed an intervention (the Pregnancy Free Club). 136

Three main themes were identified:

1. ‘Attitudes towards repeat pregnancy’, which included lifestyle and aspirations, no planning or consideration of consequences, teenage pregnancy being the norm for some, whereas some had education or career aspirations, but had problems achieving them. Others did not want to conform to stereotypes and wanted to do the best for themselves and their children. It also included rationales for repeat pregnancy which included reverse life-course rationalisation (completing their family before moving on), partners’ desire to have a baby and that it might make them stay with them, or replacing a loss (after stillbirth, abortion or a child being taken into care).

2. ‘Barriers and facilitators to effective postpartum contraceptive use’ relating to knowledge, information and misinformation (lack of knowledge about contraception and understanding of fertility, misconceptions about side effects of contraceptives, experience of side effects prompting method change or discontinuation with potential for gaps in protection, partners’ influence – positive and negative), access and continuity of care (timely contraceptive counselling together with supply of the preferred method was ideal; there were perceived barriers, such as restrictive clinic hours and requirement for a referral letter).

3. ‘Abortion and repeat abortion’ which included negative attitudes towards abortion (viewed from a moral stance and negative attitudes reinforced in school), influences on the abortion choice (partners, parents, peers community and lack of knowledge and misinformation and horror stories and fears for future fertility) and the need for pre- and post-abortion support and that services were patchy.

Service user feedback and overview of realist synthesis methods and preliminary findings

Rabeea’h Aslam presented details of our service user feedback session.

A meeting was facilitated by Flying Start, Barnardo’s and Swansea Young Families with a group of 17 young mothers aged 15–22 years. We sought their opinions on the following issues.

Overview of realist theory areas

Rabeea’h Aslam also presented an overview of realist synthesis methods and some preliminary findings. Realist review has developed from the philosophical traditions of critical realism, which question and address the perceived or assumed interrelationship between science, knowledge and reality. The realist research question is often summarised as ‘What works for whom, under what circumstances, how and why?’. The evidence mind map was presented. Preliminary middle-range theories are:

• motivation (reasons for getting pregnant)

  • the government’s view of teenage pregnancy

  • economic factors

  • family situation

  • question of unintended, unplanned, unwanted versus ambivalence.

• connectedness

  • to peer group

  • to family

  • to school

  • to the intervention itself

  • to the person delivering the intervention

  • mentorship

  • personal care.

• targeting

  • tailoring (including who delivers the intervention)

  • timing of the intervention

  • nature of the intervention

  • person focused rather than programme focused

  • collaborative decision-making.

• setting/environment

  • stigma

  • ‘norms’ (societal and family)

  • partner influence

  • education setting

  • intervention setting.

What do these findings mean to you within your work context?

Who do you think should hear these findings?

How do you think these findings should be delivered to these audiences to maximise uptake?

Is there a policy message?

Do these findings make sense to you?

How similar are they to the interventions you are delivering locally?

Outline of conversation and exercise

Go around the room and ask everyone to introduce themselves, and then either of the below: tell us their favourite TV show or ‘something (interesting?) about yourself’.

Outline of conversation

According to our review findings, if they use contraception consistently, it reduces multiple pregnancies.

1. Authors suggest wide usage of LARCs such as injection or implants in young mums. The reason they give is say that the possibility of forgetting to take the pill is increased in this age group (13–19 years). What do you think of this finding?

2. Researchers argue that pills are not a good method because it requires daily compliance and young mums fear weight gain and other side effects. What do you think of this finding?

3. Advanced supply of emergency contraception emergency contraception should be provided with primary contraception. Y/N

4. Let’s we think about the argument that increasing access to emergency contraception means that young mums use emergency contraception as a primary contraceptive method.

Exercise

Picture of each contraceptive methods on a flipchart, with space to jot down summary points keywords, etc. We will lead them through the above statements, and work accordingly.

Exercise with young mums

Using the above seven points as guidance, we will use like and dislike votes (flash cards which have the Facebook sign of like and dislike) and then speak to them in easy language about those main interventions from our study.

Outline of conversation

We’ve talked about complex intervention, now what we want to know, what things support young mothers’ decisions or choices?

Exercise

Mid, Family, Services, Opportunities (thought boxes on flipcharts, and comments below) Access to services means, when do they access, do they have access, do they know about the services? What stops from accessing?

Views on complex interventions continued

The young mothers also said that they had trouble dealing with tantrums and screaming. They agreed that someone coming to their house once or twice a month would be very helpful.

Who do you speak to when you’re going out of your mind?

She called me a bad mother because she was constipated when she went from breast feeding to normal milk.

They found there was a difference between the health visitor and a Teen Start visitor, and would prefer a person who was specialised in working with young mothers.

They also said many were seeking to continue education, then postnatal and prenatal care needs to include crèche facilities and intervention needs to occur at school. They would have liked support after a second child was born because it is hard each time.

C: completeness

We will not include any reports in the in-depth review that are incomplete, that is:

• those that, after exhaustive searches, only a partial record, such as an abstract or short report, can be obtained for

• a report of an intervention where the components of the intervention are not fully described

• a report of an evaluation that does not fully describe the methods used

• a report that does not contain data that address the review question.

A: accuracy

Accuracy is generally assessed using quality appraisal criteria. Studies will not be excluded from in-depth review on the grounds of poor quality as long as they pass the screening questions of the MMAT,54 since further appraisal may be not feasible or appropriate when the answer is ‘no’ or ‘can’t tell’ to one or both of the following screening questions:

• Are there clear qualitative and/or quantitative research questions (or objectives), or a clear mixed-methods question (or objective)?

• Do the collected data allow the research question (or objective) to be addressed? For example, consider whether or not the follow-up period is long enough for the outcome to occur (for longitudinal studies or study components).

Studies that pass the screening questions and meet the criteria for completeness, relevance and timeliness will be included for in-depth review and undergo full quality appraisal, and the quality of studies will be addressed in the discussion.

R: relevance

Stakeholders agreed that, while there were some commonalities between study populations in different countries, contextual variations placed limitations on the applicability of their findings to UK settings. Therefore, we will limit the inclusion of non-UK studies in the in-depth review to those published in peer-reviewed journals. However, since published literature from the UK in this field is scarce, we will conduct additional targeted searches to identify grey literature, which we will include in the in-depth review if it meets the criteria for completeness, accuracy and timeliness.

T: timeliness

Stakeholders commented that, because of changes in social norms, behaviours, etc., studies should be reasonably up to date. Our cut-off date for literature searches was 1995; we will not include studies where data were collected before 1990.

Motivation: reasons for getting pregnant

• Political view (i.e. the state’s view of teenage pregnancy).

• Economic.

• Family situation.

• Question of unintended, unplanned or unwanted pregnancy, or ambivalence.

The likelihood of teenage pregnancy might be affected by paying attention to ‘motivation’; in particular, the reasons why a girl is pregnant, asking whether the pregnancy was unplanned, unintended or unwanted, or whether or not the girl in question was ambivalent to motherhood and/or contraception use. Interventions that pay attention to these factors may be more effective.

The likelihood of teenage pregnancy might be affected by paying attention to ‘state and economic views of teenage pregnancy.’ The Government and the media tend to view teenage pregnancy as a negative phenomenon, as teenage motherhood increases the likelihood of financial dependence on the state, thus perpetuating the negative view; however, in certain cases, teenage pregnancy can be a positive experience that does not result in financial dependence on the state.

Connectedness

• To peer group.

• To family.

• To school.

• To the intervention itself.

• To the person delivering the intervention.

• Mentorship.

• Personal care.

• Collaborative decision-making.

The likelihood of teenage pregnancy might be affected by paying attention to ‘connectedness’: positive/good relationships with family members, peers, their school and, in certain cases, the person delivering an intervention can provide a teenage girl with a sense of belonging and attachment to a group (which could in turn lead to feelings of empowerment). Therefore, interventions that pay attention to these factors may be more effective.

Targeting

• Tailoring (including who delivers the intervention).

• Timing of the intervention.

• Nature of the intervention (e.g. single component vs. multicomponent).

• Person focused rather than programme focused.

• Collaborative decision-making.

The likelihood of teenage pregnancy might be affected by paying attention to ‘tailoring’: allowing for certain aspects of the intervention to be tailored (e.g. timing, location and intervention focus) can provide a personalised intervention that reduces barriers to attendance, increasing the likelihood of engagement and adherence to an intervention.

Setting/environment

• Stigma.

• ‘Norms’ (i.e. society norm, family norm, etc.).

• Partner influence.

• Family setting.

• Education setting.

• Intervention setting.

The likelihood of teenage pregnancy might be affected by paying attention to ‘setting/environment’: family and education settings may create a social norm in which teenage pregnancy is deemed an acceptable and a common occurrence without stigma, leading to the ‘normalisation’ of teenage pregnancy.

The likelihood of teenage pregnancy might be affected by paying attention to ‘partner influence’: a partner may have their own ideals on family size or, in cases where the teenage girl in question has given birth to a previous partner’s child, the current partner may have his own desires to build a family with the girl. Both of these scenarios could influence the teenage girl.

The likelihood of teenage pregnancy might be affected by paying attention to ‘intervention setting’: providing the intervention in settings such as the home, community or school could minimise barriers to participation and attendance, increasing the likelihood of overall intervention engagement and continued attendance.

As part of the Cardiff Task and Finish Group meeting, summary statements were prepared and presented for each theory area to explain the emerging themes of evidence to the attendees. These summary statements are shown below.

Connectedness

The likelihood of teenage pregnancy might be affected by paying attention to ‘connectedness’. Adolescents need social support from parents, partners, friends and peers to develop their self-esteem, in order to increase their self-efficacy and to reinforce their choices regarding the delay or avoidance of repeat pregnancy with regard to, for example, accessing and consistent use of contraception. Adolescents without at least two people in their lives to verbalise their importance as a person or confirm their parenting skills were less likely to use birth control consistently. Interventions that provide social support through relatable sources to the mother, such as peers, mentors or role models, can assist with filling the gap in social support. Social support that builds trusting relationships can provide an adolescent mother with assistance, guidance, and confirmation of her self-worth. Interventions that use peers, mentors and role models may be more successful if they adopt a supportive ‘big sister’ role, rather than an authoritarian ‘mother’ or ‘teacher’ role, as many adolescents have fractured relationships with their own mother or conflicts with authority (e.g. teachers or law enforcement).

Intervention content

The likelihood of teenage pregnancy might be affected by paying attention to ‘intervention content’. Because of multiple levels of influence upon adolescents, from families, peers and the community, knowledge and information of contraception is, alone, not enough and often does not lead to implementation. Some have argued for the use of social ecological theory to examine individual, dyad, family, peer/community and social system factors in order to develop intervention frameworks. The issue of adolescent pregnancy involves family, societal, medical and educational components. Therefore, interventions should be multifaceted providing a range of support and services that address the complex needs of young mums. Common components that have been consistently found in effective programmes include peer counselling, case management, school completion and career strategies, and contraceptive availability.

Intervention delivery

The likelihood of teenage pregnancy might be affected by paying attention to ‘intervention delivery’. Interventions delivered in schools encourage a strong focus on remaining in education, as the adolescent mother usually needs to be enrolled in school to access these services and interventions. Engagement in school-based activities has been found to be a protective factor against pregnancy in adolescence. This could see a shift in resources in schools from downstream, reactive interventions to more upstream, proactive interventions that aim to prevent an initial pregnancy in adolescence rather than repeat pregnancies in adolescence. The dichotomy between the access and availability of services for urban versus rural adolescents was also evident, demonstrating a need for better access in rural areas.

Perspective/context

The likelihood of teenage pregnancy might be affected by paying attention to ‘perspective/context’. Family dynamics, past experiences and expectations of gender roles can influence an adolescent girl’s perceptions of motherhood. Interventions should be sensitive to these personal experiences when developing and delivering interventions; for example, an adolescent who has experienced multiple miscarriages may experience feelings of guilt, loss and depression leading to thoughts of repeat pregnancy to replace the baby or babies that she has lost. If an intervention pays attention to this history and its potential effect on decision-making, the appropriate counselling may be delivered as a part of the intervention allowing the adolescent to deal with these feelings of loss without resulting in a repeat pregnancy.

Other goals and aspirations

The likelihood of teenage pregnancy might be affected by paying attention to ‘other goals and aspirations’. Other goals and aspirations, outside of motherhood, may help teenagers to equate consistent contraception use with obtaining the lifestyle they want. Without other alternatives, motherhood may become the only option to provide girls with an opportunity for success and autonomy. However, if the adolescent is already a mother there is a need for flexibility and additional support to help these girls achieve their goals through intervention tailoring and structure, and providing, for example, extended deadlines for school or college course work or graduation requirements. Interventions that provide a range of services, such as birth control, social support, service co-ordination, health education, life skills and employment training, may equip adolescents with the skills they need to seek out alternative life choices.

Perceptions of parental responsibility

The likelihood of teenage pregnancy might be affected by paying attention to ‘perceptions of parental responsibility’. Adolescent mothers report their desire to provide a better upbringing for their own child than they experienced themselves; they see an opportunity to rectify the mistakes made in their own childhood. There is also a difficulty for adolescent mothers in developing autonomy and self-sufficiency in parenting, as maternal grandmothers often undermine the adolescent’s self-sufficiency by taking over the parenting role. This clash for parental responsibility can lead the adolescent to a repeat pregnancy in order to gain autonomy and the sole parenting role of a child, as the grandparent is now needed to parent the older sibling(s). Interventions should aim to support self-sufficiency by equipping parents with the skills they need to be the capable, competent parents they wish to be.

Setting/environment individual

The likelihood of teenage pregnancy might be affected by paying attention to ‘setting/environment individual’. Personal characteristics and circumstances influence an adolescent girl’s ideas about the likelihood, implications and potential benefits of becoming a mother during adolescence. The personality of an adolescent is important. There are sexually active adolescents who do not become pregnant. It has been suggested that exploring the characteristics and personalities of never-pregnant, sexually active adolescents to uncover exactly why they do not get pregnant, despite sexual activity, could provide insights into effective interventions. For example, do these never-pregnant sexually active adolescents feel they can speak openly about sex with mothers, friends and partners? Do they feel the personal cost of stigma when accessing emergency contraception is less than the personal cost of a pregnancy during adolescence?

Tailoring

The likelihood of teenage pregnancy might be affected by paying attention to ‘tailoring’. There are multiple levels of influence upon an adolescent and thus interventions should be geared towards an adolescent’s readiness to change, rather than standardised messages and advice. Interventions that pay attention to adolescents’ life experiences, as well as their developmental stage, cultural context, age-appropriate impulsive and rational decision-making styles and responses to stress, may be more successful than interventions based upon a medical model. Medical models of interventions are defined as interventions that focus mainly on information and access to contraception only. An example of this type of model is the suggestion that the most effective way to help adolescent mothers might be to encourage them to use long-acting injectable or implant contraceptive methods. Interventions should also be sensitive to barriers, including practical, emotional and psychological barriers, and should overcome these barriers; for example, interventions should address practical issues of access by providing interventions at home or by providing transport if the intervention is to be delivered in another setting, such as a community centre. Interventions also need to understand adolescent sexual behaviour from the point of view of the adolescents themselves. Many interventions are designed based upon adult beliefs of adolescent sexual activity and the consequences of adolescent pregnancy, which may lack relevance to the adolescent population that the intervention aims to serve. Interventions should be modelled using adolescent principles and perceptions of sex and relationships, and the potential consequences of pregnancy during adolescence.

These findings resonated with attendees at the stakeholder group, the stakeholders only wished to add, with regards to tailoring, that they felt interventions needed to be tailored to the individual according to their circumstances at the time. They said ‘girls complain that they are not listened to by professionals’. They also stated ‘we really need to find out what they really need and want, and understand what they are asking for – there is too much generalisation, and perceptions of what girls want are not accurate’.

Excluded because of study type (usually the article did not report research with useable data) (n = 44)

Akinbami LJ, Cheng TL, Kornfeld D. A review of teen–tot programs: comprehensive clinical care for young parents and their children. Adolescence 2001;36:381–93.

Alford S, Rutledge A, Huberman B. Science and Success: Programs that Work to Prevent Subsequent Pregnancy Among Adolescent Mothers. Washington, DC: Advocates for Youth; 2009.

Bailey PE, Bruno ZV, Bezerra MF, Queiroz I, Oliveira CM, Chen-Mok M. Adolescent pregnancy 1 year later: the effect of abortion vs. motherhood in Northeast Brazil. J Adolesc Health 2001;29:223–32.

Baldwin MK, Edelman AB. The effect of long-acting reversible contraception on rapid repeat pregnancy in adolescents: a review. J Adolesc Health 2012;52:S47–53.

Barlow J. Young mothers: contraception choices and other concerns. Practice Nurse 2005;29:24.

Baytop CM. Evaluating the effectiveness of programs to improve educational attainment of unwed African American teen mothers: a meta analysis. J Negro Educ 2006;75:458–77.

Bouris A, Guilamo-Ramos V, Cherry K, Dittus P, Michael S, Gloppen K. Preventing rapid repeat births among Latina adolescents: the role of parents. Am J Public Health 2012;102:1842–7.

Childree J. Abstract and Critique for Repeat Pregnancies Among Adolescent Mothers. Abstract and research proposal for a thesis. West Florida: University of West Florida; 2007.

Collier J, Blake H. Sexual and reproductive health in pregnant teenagers presenting for antenatal care or for termination. Curr Paediatr 2006;16:211–15.

Corcoran J, Pillai VK. Effectiveness of secondary pregnancy prevention programs: a meta-analysis. Res Soc Work Pract 2007;17:5–18.

Coren C. Some teenage mothers place high priority on avoiding repeat pregnancy in their early postpartum months. Perspect Sex Reprod Health 2004;36:34–5.

Dailard C. Reviving interest in policies and programs to help teens prevent repeat births. Guttmacher Rep Public Policy 2000;3:1–11.

Edwards S. Incentives draw teenage mothers to support groups, but participation does not prevent repeat pregnancy. Fam Plann Perspect 1997;29:191–2.

Núñez-Urquiza RM, Hernández-Prado B, García-Barrios C, González D, Walker D. Embarazo no deseado en adolescentes, y utilizacion de metodos anticonceptivos posparto. Salud Publica Mexico 2003;45:S92–S102.

Florida State University Center for Prevention and Early Intervention Policy. Subsequent Pregnancies and Births Among Adolescent Mothers FACT SHEET. 2005. URL: www.cpeip.fsu.edu/resourceFiles/resourceFile_76.pdf (accessed 20 June 2014).

Furey A. Are support and parenting programmes of value for teenage parents? Who should provide them and what are the main goals? Public Health 2004;118:262–7.

Greer F, Levin-Epstein J. One Out of Every Five: Teen Mothers and Subsequent Childbearing. Washington, DC: Center for Law and Social Policy; 1998.

Klerman LV, Baker BA, Howard G. Second births among teenage mothers: program results and statistical methods. J Adolesc Health 2003;32:452–5.

Knafl K. Effectiveness of a family support center approach to adolescent mothers: repeat pregnancy and school drop-out rates. J Child Fam Nurs 1998;1:88–90.

Lopez LM, Hiller JE, Grimes DA, Chen M. Education for contraceptive use by women after childbirth. Cochrane Database Syst Rev 2012;8:CD001863.

Madden T. Long-acting removable contraceptives prevent teen pregnancy. J Adolesc Health 2013;52:255–6.

Meade CS, Ickovics JR. Systematic review of sexual risk among pregnant and mothering teens in the USA: pregnancy as an opportunity for integrated prevention of STD and repeat pregnancy. Soc Sci Med 2005;60:661–78.

Milne D, Glasier A. Preventing repeat pregnancy in adolescents. Curr Opin Obstet Gynecol 2008;20:442–6.

Mims BL. Afrocentric perspective of adolescent pregnancy in African American families: a literature review. ABNF J 1998;9:80–8.

Nunez-Urquiza RM, Hernandez-Prado B, Garcia-Barrios C, Gonzalez D, Walker D. Unwanted adolescent pregnancy and post-partum utilization of contraceptive methods. Salud Publica Mexico 2003;45:S92–S102.

Omar HA, Fowler A, McClanahan KK. Significant reduction of repeat teen pregnancy in a comprehensive young parent program. J Pediatr Adolesc Gynecol 2008;21:283–7.

Oxtoby K. The Moat House Centre for Teenage Mums in Stockport. Community Care; 2008. URL www.communitycare.co.uk/2008/06/24/the-moat-house-centre-for-teenage-mums-in-stockport/ (accessed 13 October 2015).

Perrow F. Teenagers and double buggies: preventing a second pregnancy. RCM Midwives J 2004;7:508–9.

PhysOrg.com. Repeat Pregnancies Among UK Teenagers on the Increase. University of Nottingham; 2010. URL: http://phys.org/news152204668.html (accessed 20 June 2014).

Plastino KA. Commentary on repeat pregnancy prevention self-efficacy in adolescents: associations with provider communication, provider type and depression. South Med J 2012;105:598–9.

Porter LS, Holness NA. Breaking the repeat teen pregnancy cycle. Nurs Womens Health 2011;15:368–81.

Riera KC. Living Arrangements and the Outcomes of a Teen or Young Adult Birth. PhD thesis. College Park, MD: Department of Family Science, University of Maryland; 2011.

Rigsby DC, Macones GA, Driscoll DA. Risk factors for rapid repeat pregnancy among adolescent mothers: a review of the literature. J Pediatr Adolesc Gynecol 1998;11:115–26.

Rosengard C. Confronting the intendedness of adolescent rapid repeat pregnancy. J Adolesc Health 2009;44:5–6.

Rowlands S. Social predictors of repeat adolescent pregnancy and focussed strategies. Best Pract Res Clin Obstet Gynaecol 2010;24:605–16.

Ruedinger E, Cox JE. Adolescent childbearing: consequences and interventions. Curr Opin Pediatr 2012;24:446–52.

Seitz V, Apfel NH. Effective interventions for adolescent mothers. Clin Psychol-Sci Pract 1999;6:50–66.

Smith P, Buzi R. Workforce development as a promising approach to pregnancy prevention. J Adolesc Health 2013;1:S108.

Snow T. Pregnant pause for teenage mums. Nurs Stand 2006;20:14–15.

Nursing Standard Clinical Digest. Aggression is a predictor of rapid repeat teen pregnancy: pubertal onset age and conflict management key considerations in prevention strategies. Nurs Stand 2009;23:16.

Nursing Standard Clinical Digest. Two-times teen mothers face high risks for second baby. Nurs Stand 2008;22:16.

Stevens-Simon C. A cautionary note: letter to the editor. J Adolesc Health 2003;33:322–3.

Stevens-Simon C. The Second Chance Club. J Adolesc Health 2001;29:80.

Women’s Centre of Jamaica Foundation. Programme for Adolescent Mothers. Advocates for Youth; 2012. URL: www.aed.org/en/RH/Pubs/wsp/caseStudies/JamaicaCS.htm (accessed 30 May 2013).

Excluded because of study focus (usually the focus was on primary teenage pregnancy, not repeat) (n = 33)

Alcindo José R. Novamente grávida: adolescentes com maternidades sucessivas em Rondonópolis. [MT Pregnancy Again: Consecutive Motherhood in Rondonópolis]. PhD thesis. São Paulo: Faculdade de Saúde Pública; 2007. URL: www.teses.usp.br/teses/disponiveis/6/6136/tde-11022008–222655/pt-br.php (accessed 2 June 2013).

Aracena M, Krause M, Perez C, Mendez MJ, Salvatierra L, Soto M, et al. A cost-effectiveness evaluation of a home visit program for adolescent mothers. J Health Psychol 2009;14:878–87.

Balcazar H, Peterson G, Cobas JA. Acculturation and health-related risk behaviors among Mexican American pregnant youth. Am J Health Behav 1996;20:425–33.

Beenhakker B, Becker S, Hires S, Molano Di Targiana N, Blumenthal P, Huggins G. Are partners available for post-abortion contraceptive counseling? A pilot study in a Baltimore city clinic. Contraception 2004;69:419–23.

Birch DML. The adolescent parent: a fifteen year longitudinal study of school-age mothers and their children. Int J Adolesc Med Health 1998;10:141–53.

Blank L, Baxter SK, Payne N, Guillaume LR, Pilgrim H. Systematic review and narrative synthesis of the effectiveness of contraceptive service interventions for young people, delivered in educational settings. J Pediatr Adolesc Gynecol 2010;23:341–51.

Blank L, Baxter SK, Payne N, Guillaume LR, Squires H. Systematic review and narrative synthesis of the effectiveness of contraceptive service interventions for young people, delivered in health care settings. Health Educ Res 2012;27:1102–19.

Bond L, Lavelle K, Lauby J. A comparison of the risk characteristics of ever-pregnant and never-pregnant sexually active adolescents. J HIV AIDS Prev Educ Adolesc Child 2002;5:123–37.

Bowman KG, Ruchala PL. A comparison of the postpartum learning needs of adolescent mothers and their mothers. J Obstet Gynecol Neonatal Nurs 2006;35:250–6.

Breheny M, Stephens C. Barriers to effective contraception and strategies for overcoming them among adolescent mothers. Public Health Nurs 2004;21:220–7.

Collins ME, Lemon C, Street E. A consumer view of teen living programs: teen parents’ satisfaction with program components and services. Fam Soc 2000;81:284–93.

Dallas CM. Rapid repeat pregnancy among unmarried, African American adolescent parent couples. West J Nurs Res 2013;35:177–92.

de Sousa MCR, Gomes KRO. Objective and perceived knowledge of oral contraceptive methods among adolescent mothers. Cad Saúde Pública 2009;25:645–54.

Degazon-Johnson R. A New Door Opened: a Tracer Study of the Teenage Mothers Project, Jamaica. Early Childhood Development: Practice and Reflections 13. Following Footsteps. The Hague: Bernard van Leer Foundation; 2001.

Felton GM, Parsons MA, Hassell JS. Health behavior and related factors in adolescents with a history of abortion and never-pregnant adolescents. Health Care Women Int 1998;19:37–47.

Kaufman CE, de Wet T, Stadler J. Adolescent pregnancy and parenthood in South Africa. Stud Fam Plann 2001;32:147–60.

Lee MC, Lee SH, Chou MC. Association of risk-taking behaviors with adolescent childbearing. J Formos Med Assoc 2001;100:533–8.

McGaha-Garnett V. Needs assessment for adolescent mothers: building resiliency and student success towards high school completion. In Walz GR, Bleuer J, Yep R, editors. Compelling Counseling Interventions: Celebrating VISTAS’ Fifth Anniversary. Ann Arbour, MI: Counseling Outfitters; 2008. pp. 11–20.

Meyrick J. Repeat use of contraceptive crisis services among adolescent women. J Fam Plann Reprod Health Care 2001;27:33–6.

Mims BL. Communication Patterns in African-American Families with Repeat Adolescent Pregnancy. PhD thesis. Ann Arbour, MI: University of Michigan; 1996. URL: http://0-search.ebscohost.com.unicat.bangor.ac.uk/login.aspx?direct=true&db=c8h&AN=1999063948&site=ehost-live (accessed 16 July 2013).

Molina RC, Roca CG, Zamorano JS, Araya EG. Family planning and adolescent pregnancy. Best Pract Res Clin Obstet Gynaecol 2010;24:209–22.

Moriarty Daley A, Sadler LS, Dawn Reynolds H. Tailoring clinical services to address the unique needs of adolescents from the pregnancy test to parenthood. Curr Probl Pediatr Adolesc Health Care 2013;43:71–95.

Obare F, van der Kwaak A, Birungi H. Factors associated with unintended pregnancy, poor birth outcomes and post-partum contraceptive use among HIV-positive female adolescents in Kenya. BMC Womens Health 2012;12:34.

Phipps MG, Rosengard C, Weitzen S, Meers A, Billinkoff Z. Age group differences among pregnant adolescents: sexual behavior, health habits and contraceptive use. J Pediatr Adolesc Gynecol 2008;21:9–15.

Rome ES, Rybicki LA, Durant RH. Pregnancy and other risk behaviors among adolescent girls in Ohio. J Adolesc Health 1998;22:50–5.

Ross M. Health and Social Services Provision to Pregnant and Parenting Adolescents in Southern New Brunswick. MA thesis. Fredericton, New Brunswick: University of New Brunswick; 2009.

Rotermann M. Second or subsequent births to teenagers. Health Rep Stat Canada 2007;18:39–42.

Sadler LS, Chen JY, Daley AM, Leventhal JM, Reynolds H. Reproductive care and rates of pregnancy in teenagers with negative pregnancy test results. J Adolesc Health 2006;38:222–9.

Taffa N, Omollo D, Matthews Z. Teenage pregnancy experiences in rural Kenya. Int J Adolesc Med Health 2003;15:331–40.

Tomlinson K. Repeat Pregnancy in Teenage Mothers. PhD thesis. Hull: University of Hull; 2008. URL: https://hydra.hull.ac.uk/assets/hull:1597a/content (accessed 20 June 2014).

Trivedi D, Brooks F, Graham M, Bunn F, Wentz R. Teenage Pregnancy and Parenthood: A Descriptive Mapping of Research Studies Exploring the Role of General Practitioners, Nurses, Health Visitors or Midwives; Young Fathers/Young Men; and Peers. 2012. URL: www.herts.ac.uk/fms/documents/research/health-and-human-science-research-institutes/nice_review.pdf (accessed 30 May 2013).

Ventura W, Ventura-Laveriano J, Nazario-Redondo C. Perinatal outcomes associated with subsequent pregnancy among adolescent mothers in Peru. Int J Gynecol Obstet 2012;117:56–60.

Vishneski S. Content validation of an instrument used to measure contraceptive knowledge, attitudes and usage in adolescents with a rapid repeat pregnancy. J Womens Health 2010;19:637.

Excluded because of study population (usually the study population was not clearly defined according to our criteria for ‘teenage’) (n = 22)

Akinbami LJ, Schoendorf KC, Kiely JL. Risk of preterm birth in multiparous teenagers. Arch Pediatr Adolesc Med 2000;154:1101–7.

Boardman L, Allsworth J, Phipps M, Lapane K. Risk factors for unintended versus intended rapid repeat pregnancies among adolescents. J Adolesc Health 2006;39:e1–8.

Gilliam M, Knight S, McCarthy M. Success with oral contraceptives: a pilot study. Contraception 2004;69:413–18.

Gilliam ML, Warden MM, Tapia B. Young Latinas recall contraceptive use before and after pregnancy: a focus group study. J Pediatr Adolesc Gynecol 2004;17:279–87.

Jacoby M, Gorenflo D, Wunderlich C, Eyler AE. Rapid repeat pregnancy and experiences of interpersonal violence among low-income adolescents. Am J Prev Med 1999;16:318–21.

Kanku T. Attitudes, perceptions and understanding amongst teenagers regarding teenage pregnancy sexuality and contraception in Taung. S Afr Fam Pract 2010;52:563–72.

Kelly LS, Sheeder J. Why lightning strikes twice: postpartum resumption of sexual activity during adolescence. J Pediatr Adolesc Gynecol 2005;18:327–35.

Kuroki LM, Allsworth JE, Redding CA, Blume JD, Peipert JF. Is a previous unplanned pregnancy a risk factor for a subsequent unplanned pregnancy? Am J Obstet Gynecol 2008;199:517.e1–7.

Lemay CA, Cashman SB, Elfenbein DS, Felice M. Adolescent mothers’ attitudes towards contraceptive use before and after pregnancy. J Paediatr Adolesc Gynecol 2007;20:233–40.

Matsuda Y, Masho S, McGrath JM. The relationship between repeated unintended pregnancies and current contraceptive use: National Survey of Family Growth (NSFG) 2006–2008 data. J Community Health Nurs 2012;29:163–72.

Mentula MJ, Niinimäki M, Suhonen S, Hemminki E, Gissler M, Heikinheimo O. Young age and termination of pregnancy during the second trimester are risk factors for repeat second-trimester abortion. Am J Obstet Gynecol 2010;203:107.e1–7.

Mims B, Biordi DL. Communication patterns in African-American families with adolescent mothers of single or repeat pregnancies. J Natl Black Nurses Assoc 2001;12:34–41.

Omar H, Fowler A, D’Angelo S. Improved continuation rate of depot-medroxyprogesterone acetate in adolescent mothers. Int J Adolesc Med Health 2002;14:149–52.

Peipert JF, Madden T, Allsworth JE, Secura GM. Preventing unintended pregnancies by providing no-cost contraception. Obstet Gynecol 2012;120:1291–7.

Qin J-P, Chen S, Di N, Yang Y-P, Zhou L, Zhang D-k. Evaluation of the effect of post-abortion counseling and education among unmarried abortion adolescents. Chung Hua Fu Chan Ko Tsa Chih 2010;45:201–4.

Rose SB, Cooper AJ, Baker NK, Lawton B. Attitudes toward long-acting reversible contraception among young women seeking abortion. J Womens Health 2011;20:1729–35.

Sundby J, Svanemyr J, Maehre T. Avoiding unwanted pregnancy-the role of communication, information and knowledge in the use of contraception among young Norwegian women. Patient Educ Couns 1999;38:11–19.

Tocce KM, Sheeder JL, Teal SB. Rapid repeat pregnancy in adolescence: do immediate postpartum contraceptive implants make a difference? Am J Obstet Gynecol 2012;206:e1–7.

Tocce K, Sheeder J, Python J, Teal SB. Long-acting reversible contraception in postpartum adolescents: early initiation of etonogestrel implant is superior to IUDs in the outpatient setting. J Pediatr Adolsc Gynecol 2012;25:59–63.

Torjesen I. Jinny had a real desire to make a difference. Nurs Times 2008;104:24–5.

Waggoner MR, Lanzi RG, Klerman LV. Pregnancy intentions, long-acting contraceptive use and rapid subsequent pregnancies among adolescent and adult first-time mothers. J Child Adolesc Psychiatr Nurs 2012;25:96–104.

Whitaker AK, Dude AM, Neustadt A, Gilliam ML. Correlates of use of long-acting reversible methods of contraception among adolescent and young adult women. Contraception 2010;81:299–303.

Unobtainable report (usually because a web page was no longer available) or full citation details missing (n = 6)

Baltimore Region Community Programs, Government and Community Affairs. College Park, MD: University of Maryland; 2012.

Community-based computer-assisted motivational intervention reduces repeat births to adolescents. J Natl Med Assoc 2010;102:154.

King-Marshall E. Achievement Motivation Association with Teen Birth and Repeat Birth: A Qualitative Study. Gainesville, FL: University of Florida; 2012.

Nottingham University Hospitals NHS Trust. Reducing Repeat Pregnancies: A Randomised Control Trial Of A Targeted Family Planning Intervention For Teenage Mothers & For Teenagers Who Have A Termination Of Pregnancy.

Barrett, The Honors College. Student Spotlight on Kaydi Flowers. 2012.

VioPro Database Record #3616. The Mission of Abuelas y Jovenes is to Mobilize Older Women as Abuelas to Visit Young Pregnant and Parenting Latina Women. 2010.

Studies included in the mapping exercise that did not meet the CART criteria and were excluded from the in-depth review (n = 48)

Grey literature identified in Google search that did not meet the CART criteria (n = 21)

Studies identified in update searches and excluded (n = 5)

Studies identified in citation searches and excluded (n = 3)