Background: Postpartum depression (PPD) is a debilitating illness that can have adverse consequences for a mother and her baby. Research suggests that adolescent mothers are at an increased likelihood of developing PPD, and may experience PPD differently than adult mothers. Despite this, there has been a paucity of research exploring the unique trajectory of adolescents vulnerable to developing PPD.
Aims: This study aims to identify and appraise the quality of the evidence for the correlates of PPD severity, and ascertain the potential risk factors for its development in this population.
Methodology: A systematic review will be undertaken to comprehensively search for and identify relevant studies outlining correlates of symptom severity and/or risk factors associated with adolescent PPD. If deemed appropriate, studies reporting the statistical strength of associations between correlates/risk factors and PPD will be aggregated and a meta-analysis will be conducted. The first author and an independent reviewer will screen studies, extract data, and assess the quality of each included study using specified quality appraisal tools. Findings will be summarised in a table, demonstrating the overall quality of evidence from included studies, and pooled estimates of effect, if reported. A narrative synthesis of the studies will also be carried out, structured around type of correlate/risk factor. Any discrepancies in data collection or analysis will be resolved through consensus, or involvement of a third independent reviewer. The potential implications for clinical policy and interventions will be discussed.
Adolescence is a time of profound developmental, psychological and social transition and is associated with an increased risk of developing depression. Compared to childhood, overall depression prevalence estimates double to approximately 5.6% for 13–18 year-olds, with a disproportionate increase for girls (Costello, Erkanli, & Angold, 2006; Hankin & Abramson, 2001).
A subset of adolescent girls who are especially vulnerable to depression are pregnant and childrearing adolescents. Moderate to severe depression experienced during the first year postpartum is known as postpartum depression (PPD; Beck & Gable, 2001). Symptoms may include: feelings of loneliness and insecurity; anxiety; lack of concentration; difficulty sleeping; obsessive thinking; and feelings of parental inadequacy. PPD differs from ‘baby blues’ in that symptoms persist for more than two weeks, although distinguishing between less severe symptoms of depression and the supposed “normal” course of childrearing poses a challenge to accurate identification and diagnosis of PPD (Robertson, Grace, Wallington, & Stewart, 2004). PPD can contribute to emotional, behavioural and interpersonal problems for the mother, which are likely to adversely affect her interaction with, and provision of care for, her baby (Jacobsen, 1999). Furthermore, PPD can be the precursor of chronic recurrent depression; the risk of recurrence of a depressive disorder after one postpartum episode is 25% (Wisner, Parry, & Piontek, 2002).
Adolescent PPD: A Unique Experience
Research suggests that adolescent mothers experience an increased risk of PPD compared with non-childbearing adolescents and non-adolescent mothers (Mollborn & Morningstar, 2009; Lanzi, Bert, & Jacobs, 2009). Rates of PPD among women of all ages in western societies range from 7% – 13%, yet estimates increase to 30% – 60% in adolescent mothers (Leadbeater & Linares, 1992; Cynthia Logsdon, Birkimer, Simpson, & Looney, 2005). Such statistics are alarming given the high rates of adolescent conception. In 2014, the UK had an average of 6.8 live births per 1000 girls aged 15 – 17, this is higher than the European average of six. Furthermore, Scotland has a higher rate of teenage pregnancy than most other Western European countries; the teenage pregnancy rate was 34.1 per 1,000 women in 2014 (Office for National Statistics, 2016).
Furthermore, research suggests that depressive symptoms persist in adolescent mothers for more years after delivery than adult mothers (Schmidt, Wiemann, Rickert, & Smith, 2006). This may result from the significant socio-economic disadvantage faced by younger mothers in terms of fewer educational qualifications, lower employment levels and lower income (Bradshaw, Schofield, & Maynard, 2014). Further, adolescent mothers tend to have lower infant birth weights, increased infant mortality, more unintended pregnancies, and later engagement with services compared with adult mothers.
The above highlights the inequality experienced by adolescent mothers with regards to the prevalence of PPD and other adverse perinatal outcomes. Key policy drivers in Scotland aim to reduce the inequalities in maternal and infant health outcomes, including improving early access to equitable perinatal care regardless of mother characteristics/circumstances, and reducing poor wellbeing associated with adolescent pregnancy (Maternity Services Action Group Scotland, 2011; ‘The Pregnancy and Parenthood in Young People [PPYP] Strategy’, Scottish Government, 2016). The existence of such policy frameworks demonstrates that this inequality is of national concern.
Despite these key policy drivers, and research suggesting adolescent PPD is a unique experience, there is a paucity of research into this population. More specifically, the trajectory of adolescents who go on to have PPD is not well known, including the correlates and risk factors associated with this vulnerable population. Reid and Meadows-Oliver (2007) reviewed 12 research-based articles exploring the correlates of adolescent PPD in the first year postpartum, and found that more family conflict, low self-esteem, and few social supports were all associated with increased rates of PPD, although there was disagreement among results, and the studies varied in their quality and sample characteristics. Furthermore, they did not include unpublished literature in their review, which may have led to publication bias. These factors may have limited the usefulness and applicability of the conclusions drawn. A more recent review explored the correlates and consequences of adolescent PPD in 23 studies (Kleiber & Dimidjian, 2014). They found that interpersonal factors, parenting/child factors, SES, self-esteem, coping, and stress may contribute to adolescent PPD. The authors also noted that within the adolescent population, the risk factors identified to predispose someone to PPD, may also have predisposed the person to adolescent pregnancy; this presents a potential barrier to the accurate identification of independent risk factors for adolescent PPD. Only studies comparing adolescent mothers who have PPD against those who do not, will be able to delineate the specific risk factors associated with adolescent PPD. Furthermore, it is also imperative to consider the normative course of adolescence itself, as becoming a mother during early, as opposed to late, adolescence will likely confer different risk factors. Kleiber and Dimidjian (2014) did not systematically evaluate the quality of their reviewed studies, nor state the degree to which they addressed the aforementioned barriers, potentially limiting the strength of their conclusions.
To date, there has been no systematic review of literature outlining the correlates of severity of adolescent PPD, nor the specific risk factors within this population. Given the high rate of teenage births in the UK, particularly in Scotland, and the increased vulnerability of developing PPD in adolescence compared to adulthood, research into this population is of paramount importance.
This review will address the current knowledge gap regarding correlates and risk factors associated with PPD in adolescent mothers. A rigorous review and appraisal of the current adolescent PPD risk factor literature will be important in helping to inform the evidence base and help to attain current policy aims e.g. those outlined in the PPYP Strategy (Scottish Government, 2016), as well as influencing future practice and policy aimed at this population. Improved knowledge could lead to early identification of this debilitating illness, as well as recommendations for efficient and accurate antenatal and postnatal screening for vulnerable adolescents. It may also benefit clinical and/or public health interventions, such as preventative interventions aimed at adolescent mothers with varying PPD symptom severity, and contribute to the understanding of potential barriers to help-seeking within this population.
Furthermore, the proposed review will likely highlight what knowledge is lacking within the literature, and will therefore aim to guide future research efforts.
- What are the correlates associated with severity of PPD in adolescent mothers?
- What are the risk factors associated with PPD in adolescent mothers?
- What is the quality of the evidence for these correlates and/or risk factors?
Furthermore, if a meta-analysis is undertaken, the following research question shall be asked:
What is the strength of the relationship between specific correlates and/or risk factors and occurrence of PPD in adolescent mothers?
- To identify the correlates that are associated with severity of PPD in adolescent mothers.
- To identify the risk factors for developing PPD in adolescent mothers.
- To appraise the quality of the existing evidence regarding correlates and risk factors associated with adolescent PPD, and to make recommendations for future research.
To identify, if appropriate, the strength of the relationship between specific correlates and/or risk factors and occurrence of PPD in adolescent mothers.
A systematic review will be undertaken to identify relevant studies outlining correlates/risk factors associated with adolescent PPD. If appropriate, studies reporting the statistical strength of associations between correlates/risk factors and PPD will be aggregated and a meta-analysis will be conducted to determine a quantitative estimate of effect size (Field & Gillett, 2010).
A systematic literature review aims to employ a rigorous, transparent and objective scientific methodology to summarise the relevant evidence in a research area. It appraises the quality and potential bias in study designs, collection, analysis, interpretation or publication of data (CRD, 2009). Systematic reviews are easily replicable, more reliable than other literature reviewing methods, and are gaining credence in influencing policy-makers and clinical opinion (Petticrew & Roberts, 2006; Tacconelli, 2010).
This review will follow the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA; Moher, Liberati, Tetzlaff, Altman, & PRISMA Group, 2009), and if appropriate, the proposed reporting for Meta-Analysis of Observational Studies in Epidemiology (MOOSE) guidelines (Stroup et al., 2000)
General Inclusion/Exclusion criteria:
- It is expected that most included studies will have used observational designs (specifically case-control or cohort designs), although experimental designs will also be considered (e.g. a trial of a preventative intervention targeting putative correlates/risk factors for PPD). Qualitative studies, case-study or case-series designs will be excluded from this review.
- Investigated and reported at least one correlate/risk factor associated with maternal adolescent PPD.
- PPD will be defined using Beck & Gable’s (2001) definition; onset of the illness must be within one year of childbirth, but assessed after two weeks postpartum to avoid reporting ‘baby blues’.
- Carried out between 1990-2017, as Major Depressive Disorder with postpartum-onset was not recognised in the ICD-10 until 1992, and likewise the DSM-IV in 1994.
- Specified a validated measure of PPD (assessed by clinical interview or self-report).
- Written in the English Language, or have English translation available.
- Age of participants must be 19 years or younger. If a study includes a sample with a larger age range, the study will be included if subgroup analyses have been conducted on adolescents.
Additional criteria for RQ ii):
- Studies must have compared risk factors for PPD in adolescent mothers with and without PPD.
Additional criteria for RQ iv):
- Correlates/risk factors identified and measured in two or more studies.
- Studies have reported an appropriate effect size for correlates/risk factors associated with PPD (e.g. Pearson’s correlation coefficient, r, Cohen’s d, or odds ratios, OR), and corresponding 95% confidence intervals, or enough data is reported to allow for an effect size calculation.
Literature Search Method
Through consultation with a specialist librarian at Edinburgh University, a sophisticated search strategy will be employed (See Appendix 6.3). Published and unpublished literature will be examined through searching of reference lists of relevant articles or reviews; hand-searching of relevant journals, and an author search on known authors who have conducted research on the review topic.
The following databases relating to psychological, medical and social science literature will be searched: PsycINFO; EMBASE; MEDLINE; ASSIA; CINAHL; MIDIRS. Grey literature will be searched due to the possibility of reported over-estimates of effect size resulting from publication bias (unpublished studies are less likely to have statistically significant results compared with published studies (Song et al., 2010). Unpublished research will be sought by searching the following databases: ProQuest Dissertations & Theses Database; OpenGrey; GreyNet. Authors of conference abstracts will be contacted for full manuscripts of articles. Additionally, internet searches will be carried out, using search engines e.g. Google Scholar.
Study titles and abstracts will be assessed for relevance, and full copies of manuscripts for studies likely to meet selection criteria will be obtained and screened to identify any available sister papers. Final included studies will be reviewed independently by two reviewers. Disagreements will be resolved through consensus or a third independent reviewer.
In accordance with the quality guidelines for reporting systematic reviews (Moher et al., 2009), the PRISMA flow diagram will be used to report the search process (See Appendix 6.4). This flow-chart provides information on the number of identified, excluded and included studies, as well as explanations for exclusions.
Data Extraction Form
As recommended by Moher and colleagues (2009), a data extraction form has been created and tailored for use in this review (See Appendix 6.5). Use of a data collection form provides consistency and reduces risk of bias in data collection (Higgins & Green, 2011).
Quality Appraisal Tools
The quality and risk of bias for each included study will be determined through use of a quality appraisal tool. Separate tools will be used for studies using an experimental, cohort or case study design, as recommended by SIGN 50 (SIGN, 2015). The SIGN checklists aim to balance methodological rigor with practicality of use. The quality assessment tool will rate the studies across the following: selection bias, study design, potential confounders, data collection/analysis methods, measurement of correlates/risk factors, withdrawal and dropout rates, and reporting of results (See Appendix 6.6.1– 6.6.3).
For observational studies, criteria will be assessed as: Well Covered/Adequately Addressed/Poorly Addressed/Not Addressed/Not Reported/Not Applicable. For experimental studies, criteria will be assessed as: Yes/No/Can’t Say. Overall quality of the studies will consist of a rating of ‘High Quality’ (++) if all or most of the criteria have been fulfilled, ‘Acceptable’ (+) if some have been fulfilled, and ‘Low Quality‘ (-) if few or none have been fulfilled.
Results from the databases will be collated using Mendeley Version 1.17 to facilitate the removal of duplicates. The first author and an independent reviewer will screen studies and extract data using two excel data spreadsheets. The data extraction tool will initially be piloted on a subset of studies and amended as required. A reason will be provided for the exclusion of all full-text papers screened. The same reviewers will assess the quality of each included study using the quality appraisal tool. Any discrepancies will be resolved through consensus, or involvement of a third independent reviewer. Forms will be completed electronically to reduce risks of error in data transcription as recommended by NHS Centre for Reviews and Dissemination (CRD, 2009).
Clinical and methodological heterogeneity of studies will be assessed to ensure sufficient homogeneity before conducting a meta-analysis. This will be done through consideration of: population; correlate/risk factor description; depression measure; study design; procedure; and reported analysis. Statistical heterogeneity will be assessed using the I² statistic, regarded as ‘significant’ if greater than 50% (Huedo-Medina, Sánchez-Meca, Marín-Martínez, & Botella, 2006). Data will be pooled if the I² statistic is less than 85% and study characteristics are similar. A random-effects model will be used unless the I² statistic is less than 25%, based on recommendations from Field and Gillett (2010). Sensitivity analyses may be conducted to assess whether variations in clinical/methodological factors affected estimation of effects.
If a meta-analysis is carried out, quantitative data synthesis will be carried out using Review Manager Version 5.2.6 software for Windows; data will be summarised using risk ratios for dichotomous outcomes, or Cohen’s d/Pearson’s r for continuous outcomes, with 95% confidence intervals. Results of the meta-analysis will be assessed visually through Forrest plots. If 10 or more studies are included in the review, a funnel plot will be used to determine the relationship between study size and effect power of studies (Egger, Smith, Schneider, & Minder, 1997). A narrative synthesis of the studies will be carried out (including those not appropriate for meta-analysis), structured around type of correlate/risk factor. A table will be used to summarise findings, demonstrating the overall quality of evidence from included studies, and pooled estimates of effect. The strengths and weaknesses of the evidence will be reported, and the relationships between the studies, including discrepancies, will be explored (CRD, 2009).
Reviewer’s bias will be minimised through having two independent assessors review the included studies. Any relevant study information deemed to be missing or require clarification, will be sought through contacting the corresponding study author(s).
An Edinburgh University Research Ethics Application Level 1 Self-Audit Form will be submitted along with this research proposal (See Appendix 6.7). No further ethical approval is estimated.
The below Gantt Chart outlines the project timescale based on 8 hours per week, including supervision.
This proposal will be retrospectively uploaded to the PROSPERO database, as recommended by the CRD (2009). The review will be drafted and submitted to a relevant peer-reviewed journal e.g. The Journal of Clinical Child and Adolescent Psychology. The findings of this research will be disseminated to relevant forums, such as the Family Nurse Partnership (FNP), and it is hoped that presentation of the findings at relevant conferences will be undertaken.
From a clinical perspective, the findings arising from this systematic review will add to the current knowledge base through identification of correlates/risk-factors, and potentially their relative influence on adolescent PPD. This may impact practice and research priorities within the field, and help improve interventions for vulnerable adolescent mothers potentially at risk of developing PPD.