Are Maternal Decision Aids sustained in the centre where they had been found to be effective?

Patient and public involvement

Patients and the public were not involved in this study. The study was a desk-based follow-up of published decision aids that used published literature and correspondence with the original study investigators; no patients were recruited and no patient-level data were collected. Patients and the public were therefore not involved in developing the research question, in the design or conduct of the study, in the choice of outcome measures (retrievability and continued clinical use of each decision aid), or in recruitment, as there were no participants to recruit. Patients and the public were not involved in choosing the methods or agreeing plans for dissemination. The findings, together with the plain language summary, will be shared with patient and public contributors through the NIHR Midlands Patient Safety Research Collaboration.

Study selection

We based our study on Cockburn’s previous systematic review of decision aids in maternal health.⁵ Cockburn included a wide range of decision aids, including prompts/action suggestions aimed at clinicians. We removed these clinician-facing aids, retaining only those designed to support service user choice between alternative courses of action.

Data extraction

For each included study, basic descriptive information about the decision aid was extracted. This information included the study design, clinical context, the tool’s format (digital application, computer interface, or paper), and the intended user group. The findings were positive, confirming the effectiveness of DAs, in all five cases ( Table 1).

Assessing sustainability

The assessment of sustainability was conducted in two stages. First, we assessed retrievability by reviewing the original publication and supplementary files and by searching online for the named decision aid (or an updated version). Second, we assessed continued use in routine practice through direct correspondence with the corresponding or other study authors. Email addresses were located either via the published paper itself, or by using an online search engine (Google) to locate contact information about the research team, for example via the website of the individual’s current university employment. Authors were asked whether the decision aid remained in use, whether it was still accessible, and, if it had been discontinued, the reasons for discontinuation. Follow-up emails were sent where required, with up to three contact attempts made initially. For authors who did not respond to these attempts, further follow-up was undertaken by the research team. Author correspondence was ultimately obtained for all five included decision aids. Where an author response was not initially obtained, we also sought information from the clinical service in which the decision aid had originally been evaluated, using email contacts with clinic staff. Because respondents were investigators associated with the original evaluations rather than designated organisational representatives, their statements about current organisational use were treated as respondent-reported information rather than independently verified organisational data.

Barriers to sustainability

Reasons for continuation or discontinuation were extracted directly from the correspondence and tabulated by decision aid. We used a structured, deductive descriptive approach rather than a qualitative thematic analysis. Reported reasons were first retained in their original substantive form (for example, lack of ongoing funding, licensing restrictions, lack of ownership, or the need to revise an outdated prediction model) and were then mapped to the most relevant domain of the Non-adoption, Abandonment, Scale-up, Spread and Sustainability (NASSS) framework. The mapping was used as an organising framework and was not intended to imply causal attribution beyond what respondents reported. Where a reason could plausibly relate to more than one NASSS domain, its placement reflected the principal mechanism described in the correspondence.

Ethical consideration

No patient-level data were used in this study. The work relied on published literature and professional correspondence with study authors. All communication was carried out using publicly available contact details, and no sensitive or personal information was collected.

Included Studies

Five decision aids (DAs) met our inclusion criteria ( Table 1). These patient-directed DAs spanned: mode of birth after a previous caesarean section (three DAs), thromboprophylaxis during pregnancy (one DA), and preterm labour management (one DA). Geographically, tools originated from the United States (two), the United Kingdom (one), the Netherlands (one), and a Canada-Spain collaboration (one). Publication dates ranged from 2007 to 2023, spanning sixteen years of DA development in maternity care. Three of the studies used randomised designs. Outcomes were of three types: (i) decision makers’ knowledge, (ii) overall satisfaction, and (iii) satisfaction with the decision made. The outcomes measured across the studies were uniformly positive ( Table 1).

Retrievability of decision aids ( Table 1)

Only one of the five decision aids was retrievable electronically⁹ while another was retrieved by contacting the author.⁷ The remaining decision aids could not be retrieved. The PROCEED decision aid⁶ was available only as screenshots within supplementary material and could not be retrieved as a usable tool.

Sustainability of DA in clinical use

Based on the information available at the time of data extraction, one of the five decision aids was reported to remain in routine clinical use: the Dutch tool evaluated by Vankan et al.,⁹ which an original study author reported had been implemented nationally across more than 60 hospitals in the Netherlands. The same respondent noted that an unspecified number of clinics were no longer using the tool and that the prediction model underpinning the tool required revision which had not been implemented because of financial constraints and lack of clarity over ownership. For the other four decision aids, original study authors or local clinic staff reported that the tools were no longer in use.

Patterns of withdrawal

Respondent-reported reasons for discontinuation or threatened sustainability included absence of ongoing funding, licensing and governance barriers, lack of clear ownership or stewardship, and difficulty maintaining or updating tools when underlying prediction models or clinical expectations changed ( Table 2). These categories represent a deductive mapping of the reasons stated in correspondence to NASSS domains; they should not be interpreted as independently established causes of discontinuation.

Strengths and limitations

A strength of this evaluation is that direct correspondence with investigators provided information about the post-study status of decision aids that was not consistently available in publications or supplementary materials. However, this source of evidence also requires caution. Study authors were selected because they had direct knowledge of the original decision aid and its evaluation, but they were not necessarily responsible for subsequent implementation or formally authorised to speak for the organisations in which the tools had been used. Their responses therefore provide informed investigator accounts rather than independently verified organisational assessments. We have accordingly distinguished reported status and reported reasons from verified facts throughout the manuscript.

Several limitations merit acknowledgement. First, this evaluation is limited to published, identifiable DAs with traceable authors; tools developed and used outside research contexts, including commercial products, locally produced patient information sheets, or tools embedded in electronic health record systems, are not captured. Second, the findings are concentrated in high-income settings and cannot characterise sustainability in low- and middle-income countries. Third, continued use and reasons for discontinuation were primarily established through investigator correspondence. Although author correspondence was obtained for all five included decision aids, respondents were not systematically identified as organisational implementation leads, service managers, or current tool owners. Consequently, they may not have had complete knowledge of use across all original sites, and we did not independently verify their accounts with every organisation in which a decision aid had been implemented. Fourth, the reasons for withdrawal were collected through focused correspondence rather than prospective interviews or a formal qualitative study; the NASSS classification should therefore be regarded as a deductive descriptive mapping of reported reasons, not as a comprehensive analysis of causal mechanisms. Fifth, the evaluation provides a cross-sectional snapshot, and tool availability may change over time.