Opportunities and challenges of meta-reviews encountered whilst conducing a meta-review of ‘Optimal Interventions for Perinatal Anxiety (PNA) in Primary Care’.

Conference:

SAPC ASM 2022 - UCLan

Talk Code:

3D.3

Presenter:

Victoria Silverwood

Twitter:

@v_silverwood

Co-authors:

Laurna Bullock, Shoba Dawson, Carolyn A. Chew-Graham, Tom Kingstone, Katrina Turner, Joanne Jordan.

Author institutions:

VS, LB, JJ, TK, CC-G – School of Medicine, Keele University KT, SD – Centre of Academic Primary Health Care, Bristol University TK, CC-G – Midlands Partnership NHS Foundation Trust CC-G - WM ARC

Problem

Systematic reviews of reviews, or ‘Meta-reviews’, which provide summaries of existing evidence around a specific topic, are considered the highest level of evidence on the evidence pyramid. Although the general methods used for a meta-review are often the same as those used for a systematic review, there are some important differences and unique methodological challenges that authors might encounter. This presentation will discuss some of these differences and challenges, whilst also summarising meta-review methods and how they add value to the existing literature.

Approach

A meta-review is currently being undertaken to describe the evidence base for non-pharmacological interventions used to manage women with perinatal anxiety (PNA) in primary care. A number of systematic reviews exist around this topic area and a meta-review approach will provide a synthesis of their results and conclusions. The meta-review is registered on PROSPERO (ID: CRD42021202611) and will be reported in line with PRISMA 2020 guidelines.A search strategy was developed with input from an information specialist to identify systematic reviews of interventions for PNA in twelve medical databases. Titles and abstracts were independently screened by two authors (100% by VS, 20% by LB) and full texts were fully screened and read by VS and LB. Data were extracted at the review level by both authors and compared. Discrepancies were resolved by discussion. The quality of evidence of the included systematic reviews was assessed by the AMSTAR2 tool. Data have been presented narratively and in tabular format.

Findings

38 reviews were included in the meta-review. The first challenge encountered was that there was significant heterogeneity within the included studies regarding study design and outcome measures. Whilst a narrative synthesis has been performed, this heterogeneity meant that a meta-analysis was not possible Secondly, as anticipated, there was significant overlap between the eligible systematic reviews in terms of including the same primary studies. To avoid potential bias, it is important to accurately report where this overlap occurs. Methods to address this overlap included developing a citation matrix, identifying primary studies that were included in more than one review with overlap being quantified at the review level. Thirdly, some of the systematic reviews reported results using outcomes of interest that applied to broader perinatal mental health problems, rather than PNA. The original protocol was amended to reflect this so only outcomes of interest that focused on PNA were extracted.

Consequences

Meta-reviews provide high-level evidence and are an appropriate method for summarising the evidence base when several systematic reviews already exist on the topic. The resulting overview can be useful for clinical policy decision makers. However, they can be complex and more challenging to conduct than a systematic review, so may require more time and methodological expertise.

Submitted by:

Victoria Silverwood

Funding acknowledgement:

VS is a Wellcome Trust funded Clinical PhD Fellow. This work forms part of a PhD entitled: ‘Defining optimal interventions for Perinatal Anxiety (PNA) in a primary care population: a multi-methods study.’